# PolicyPeek > Free checker that reads AI tools' own privacy policies and terms and says, in one badge, whether they train on your data: Doesn't train on your data / Trains by default, you can opt out / Trains on your data, no opt-out / Policy unclear. 66 tools, each with the quoted clause, source URL and last-checked date. Not legal advice. Current counts: 17 doesn't train, 26 trains by default with an opt-out, 12 trains with no opt-out, 6 policy unclear, 4 company-wide policy only. ## Core - [All tools and verdicts](https://getpolicypeek.com/tools/): every checked AI tool with its badge - [Methodology](https://getpolicypeek.com/about/): how verdicts are made (AI extraction + fixed rules + human review) - [Check a tool](https://getpolicypeek.com/): search or request a new check ## Questions - [All questions](https://getpolicypeek.com/questions/): dated answers per tool, by plan, with quoted sources - [Does Claude train on your data?](https://getpolicypeek.com/questions/does-claude-train-on-your-data/) - [Does Notion AI train on your data?](https://getpolicypeek.com/questions/does-notion-ai-train-on-your-data/) - [Does Perplexity AI train on your data?](https://getpolicypeek.com/questions/does-perplexity-ai-train-on-your-data/) ## Guides - [Do UK Small Businesses Need an AI Policy? (2026)](https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy.md): No UK law requires a standalone AI policy, but UK GDPR and the EU AI Act's AI literacy duty can apply. What a small-business AI policy should cover. - [EU AI Act Article 4 AI Literacy: What a 10-Person UK Business Must Do](https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business.md): What EU AI Act Article 4 (AI literacy) means for a 10-person UK business using ChatGPT: who it applies to, enforcement from 2 Aug 2026, and six practical steps. - [Is It Safe to Put Customer Data into ChatGPT? A UK GDPR Guide for Small Businesses](https://getpolicypeek.com/guides/is-it-safe-to-put-customer-data-into-chatgpt-uk-gdpr.md): Can a UK small business paste customer data into ChatGPT? What UK GDPR and the ICO say, how plans differ on training, and what to do if it happened. - [AI Acceptable Use Policy Template for a Small Business (UK, 2026)](https://getpolicypeek.com/guides/ai-acceptable-use-policy-template-small-business-uk.md): A free AI acceptable use policy outline for UK small businesses: approved tools, what staff must never enter, checking output and training. Checked Oct 2026. - [How to Train Staff on AI: A 30-Minute Session for Small Teams](https://getpolicypeek.com/guides/how-to-train-staff-on-ai-30-minute-session.md): A ready-to-run 30-minute AI training session for small teams, with agenda, hallucination demo and training record. Fits the EU AI Act Article 4 duty. - [Free AI Privacy Policy Checker: Does This Tool Train on My Data?](https://getpolicypeek.com/guides/free-ai-privacy-policy-checker.md): PolicyPeek is a free checker that reads AI tools' own privacy policies and terms and tells you, in one badge, whether they train on your data. 61 tools checked. - [How to Check If an App Trains AI on Your Data (2026)](https://getpolicypeek.com/guides/how-to-check-if-an-app-trains-ai-on-your-data.md): Find out in two minutes whether an app uses your data to train AI: where to look in its policy, the exact phrases that matter, and a free checker. - [ChatGPT vs Claude: Which Uses Your Chats for Training?](https://getpolicypeek.com/guides/chatgpt-vs-claude-training.md): ChatGPT vs Claude on AI training: both train on your chats by default and both let you opt out. Exact policy quotes, sources and check dates. - [Does ChatGPT Train on Your Data? What OpenAI's Own Policy Says](https://getpolicypeek.com/guides/does-chatgpt-train-on-your-data.md): Does ChatGPT train on your data? Yes, by default, but OpenAI's own policy says you can opt out. Here's the exact wording, plus how other AI tools compare. - [Which AI Tools Don't Train on Your Data? A UK Guide (November 2026)](https://getpolicypeek.com/guides/ai-tools-that-dont-train-on-your-data-uk.md): 15 AI tools whose own privacy policies say they don't train on your data, from Abacus.AI to Zoom, plus how ChatGPT and Claude compare. Checked Nov 2026. - [AI Tools That Train on Your Data With No Opt-Out (2026 List)](https://getpolicypeek.com/guides/ai-tools-that-train-on-your-data-no-opt-out.md): 12 AI tools whose own policies let them train on your content with no confirmed training opt-out, from Ahrefs to Suno. Quotes and sources, checked Nov 2026. - [Does Mailchimp Use Your Data for AI Training? Plus HubSpot, Ahrefs and Other Marketing Tools](https://getpolicypeek.com/guides/does-mailchimp-use-your-data-for-ai-training.md): PolicyPeek rates Mailchimp Policy unclear: the published pages do not clearly say whether your data is used for training. How HubSpot and Ahrefs compare. - [Do AI Meeting Note-Takers Train on Your Recordings? Zoom, Fireflies, tl;dv, Granola and Otter](https://getpolicypeek.com/guides/ai-meeting-note-takers-that-dont-train-on-your-data.md): Which AI note-takers train on your recordings and how to opt out. Zoom, Fireflies and tl;dv say they don't; Granola and Otter let you opt out. - [Do AI Coding Assistants Train on Your Code? Cursor, Copilot, Tabnine, Windsurf and App Builders](https://getpolicypeek.com/guides/do-ai-coding-assistants-train-on-your-code.md): Which AI coding tools train on your code: Tabnine and Cody say no, Cursor, Copilot and Windsurf let you opt out, Base44 has none. - [How to Opt Out of AI Training: Image, Video, Voice and Music Tools (2026)](https://getpolicypeek.com/guides/how-to-opt-out-of-ai-training-image-video-music-tools.md): Which AI image, video, voice and music tools train on your uploads, which let you opt out, and how. Suno, Pika, ElevenLabs, Photoroom and more. --- Source: https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy/ # Do UK Small Businesses Need an AI Policy? (2026) Short answer: there is no UK law that says "every small business must have an AI policy". But if your staff use ChatGPT, Copilot, Gemini or an AI note-taker for work, a short written policy is the simplest way to show you've thought about where your data goes, and if you have customers or staff in the EU, the EU AI Act's AI literacy duty may already apply to you. *Last checked: 24 October 2026. Based on the EU AI Act text, the European Commission's AI literacy Q&A and AI tools' own published policies. Not legal advice.* ## Is an AI policy a legal requirement in the UK? Not as a standalone document. The UK has no equivalent of the EU AI Act. But the rules you already follow still apply when staff use AI tools: if someone pastes customer or employee personal data into an AI tool, that is processing personal data under UK GDPR, and confidentiality duties in your contracts don't stop at the chat box. A policy is how you tell staff what is and isn't allowed before that happens. ## When does the EU AI Act reach a UK business? The European Commission says the AI Act applies to organisations inside and outside the EU "as long as the AI system is placed on the Union market, used in the Union or its use has an impact on people located in the EU". That explicitly includes the AI literacy duty in Article 4. So a UK business with EU-based staff, or that uses AI in work affecting EU customers, should take it seriously. ([European Commission AI literacy Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) Article 4 has applied since 2 February 2025, and national authorities supervise and enforce it from 2 August 2026. For what it requires in practice, see [EU AI Act Article 4: what a 10-person UK business must do](https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business/). ## What is the real risk without one? The everyday risk is data going somewhere you didn't expect. Many AI tools' own policies allow training on what you type unless you opt out, and some give no opt-out at all. PolicyPeek's ratings of 61 AI tools' own policies (4 November 2026) found 21 that train by default with an opt-out and 14 that train with no opt-out found. ([PolicyPeek](https://getpolicypeek.com/tools/)) Without a policy, each member of staff decides alone which tool to use and what to paste in. ## What should a small-business AI policy cover? Keep it to one or two pages: 1. **Approved tools.** Which AI tools staff may use for work, and on which plan or account (business plans often have different data terms from free ones). 2. **What never goes in.** Customer personal data, staff records, passwords, unreleased financials, anything under an NDA. 3. **Training settings.** Who switches off "use my data to improve the model" where a tool allows it. Check each tool on [PolicyPeek](https://getpolicypeek.com/). 4. **Checking output.** AI can be confidently wrong. A person checks anything that goes to a customer. 5. **Who to tell.** What staff do if the wrong data goes in by mistake. 6. **Training.** A short session so everyone knows the rules, with a record of who attended. ## How long does it take? For a small team, writing a first version is an afternoon's work, especially if you start from a template. Review it when you adopt a new tool or a tool changes its terms. ## Changes - 4 November 2026: PolicyPeek tool counts updated to 61 rated tools. The 10 tools the 28 October audit had flagged stay on the site with their existing verdicts while PolicyPeek re-checks them. ## FAQ ### Is an AI policy legally required for UK small businesses? No UK law requires a standalone AI policy. But UK GDPR applies if staff put personal data into AI tools, and the EU AI Act's AI literacy duty can apply to UK businesses whose AI use reaches people in the EU. ### Does the EU AI Act apply to UK companies? It can. The European Commission says it applies to organisations outside the EU when an AI system is used in the EU or its use affects people located in the EU. ### What should a basic AI policy include? Approved tools, what data must never be entered, training opt-out settings, human checking of AI output, who to tell about mistakes, and short staff training. ### Do I need a certificate to show AI literacy? No. The European Commission says no certificate is needed; an internal record of training and guidance is enough. --- Source: https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business/ # EU AI Act Article 4 AI Literacy: What a 10-Person UK Business Must Do Article 4 of the EU AI Act says businesses that use AI systems must take measures to support the AI literacy of their staff. For a small team using everyday tools like ChatGPT, that means: know which AI tools you use, tell staff about the risks, give them short training suited to their role, and keep a simple record. No certificate, no AI officer, no specific test. *Last checked: 24 October 2026. Based on the EU AI Act (Regulation (EU) 2024/1689) as amended by the Digital Omnibus on AI, and the European Commission's AI literacy Q&A. Not legal advice.* ## What does Article 4 actually say? Article 4(1) requires "providers and deployers of AI systems" to "take measures to support the development of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf", taking into account their technical knowledge, experience, education and training and the context the AI is used in. Following the Digital Omnibus on AI, which entered into force in mid-July 2026, the text says this "does not require providers or deployers to guarantee any specific level of AI literacy of any individual". ([Article 4](https://artificialintelligenceact.eu/article/4/), [European Commission Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) If your business uses AI tools at work, you are a "deployer". If you build AI products, you are a "provider". ## Does it apply to a UK business? It can. The Commission says the AI Act applies to organisations outside the EU when the AI system is used in the EU or its use affects people located in the EU, and that this is "also valid for Article 4". A UK business with staff in the EU, or using AI in work for EU customers, should assume it applies. ([European Commission Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) ## When does it apply? Article 4 has applied since 2 February 2025. National market surveillance authorities supervise and enforce it from 2 August 2026, and the Commission says sanctions must be proportionate. ## Does using ChatGPT count? Yes. The Commission's Q&A answers this directly: a company whose employees use ChatGPT for writing advertisement text or translating should comply, and staff "should be informed about the specific risks, for example hallucination". ## What should a 10-person business do? The Commission's Q&A lists the steps to consider. For a small team: 1. **List your AI tools.** What AI does the business use, and who uses it? Include note-takers and AI features inside tools you already pay for. 2. **Know your role.** Are you only using AI tools (deployer), or building them (provider)? 3. **Look at the risks.** What could go wrong with each tool: wrong answers, personal data going in, training on your content. Check each tool's data terms on [PolicyPeek](https://getpolicypeek.com/). 4. **Train to the role.** A 30-60 minute session covering what AI is, your approved tools, what not to paste in and how to check output. People who use AI more heavily may need more. 5. **Write down the rules.** A short AI policy gives the training something to point to. See [Do UK small businesses need an AI policy?](https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy/) 6. **Keep a record.** Who was trained, when, and on what. ## What don't you need? According to the Commission: no certificate ("an internal record of trainings and/or other guiding initiatives" is enough), no obligation to test staff knowledge, and no AI officer or governance board. ## FAQ ### Does Article 4 apply to small businesses? Yes. It applies to providers and deployers of AI systems regardless of size, though the Commission and Member States are required to support SMEs in particular. ### Is AI training mandatory under Article 4? There's no fixed mandatory course. The Commission says there's no one-size-fits-all format, but relying only on the tool's instructions is often ineffective, so short role-based training is the usual approach. ### Do I need to certify my staff's AI literacy? No. The Commission says no certificate is needed; keep an internal record of training and guidance. ### When is Article 4 enforced? It has applied since 2 February 2025; national authorities supervise and enforce it from 2 August 2026. --- Source: https://getpolicypeek.com/guides/is-it-safe-to-put-customer-data-into-chatgpt-uk-gdpr/ # Is It Safe to Put Customer Data into ChatGPT? A UK GDPR Guide for Small Businesses Short answer: not on a free or personal ChatGPT account without thinking it through first. Customer names, emails and order details are personal data under UK GDPR, so you need a lawful basis, the right account type and clear rules before staff paste them in. On a ChatGPT business plan with a data processing agreement, the picture is different. *Last checked: 24 October 2026. Based on ICO guidance, OpenAI's own published policies and PolicyPeek's checks of AI tools' policies. Not legal advice.* ## Does UK GDPR apply when I paste customer data into ChatGPT? Yes. The ICO says personal data is information that relates to or identifies a person, such as names, email addresses, phone numbers and payment information, and that the rules cover businesses of every size, from sole traders up. ([ICO advice for small organisations](https://ico.org.uk/for-organisations/advice-for-small-organisations/getting-started-with-gdpr/data-protection-advice-on-common-topics)) Putting that data into an AI tool is processing it. ## What lawful basis do I need? The ICO has answered this for a business using generative AI to draft replies to clients. It says legitimate interests can work if you can show the AI use is necessary and proportionate and doesn't override your clients' rights. Consent can also work where you have a direct relationship with the people concerned. If the data includes special category data, such as health information, you also need a separate Article 9 condition, for example explicit consent. ([ICO previously asked questions](https://ico.org.uk/for-organisations/advice-and-services/innovation-advice/previously-asked-questions/)) The ICO also says you must decide and document your lawful basis before you start, and include it in your privacy notice. ([ICO: lawfulness in AI](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/guidance-on-ai-and-data-protection/how-do-we-ensure-lawfulness-in-ai/)) ## Does ChatGPT train on customer data I paste in? It depends on the account: - **Free and personal ChatGPT:** OpenAI's help centre says "When you use our services for individuals (Free, Plus, Pro), we may use your content to train our models." and "To opt out, turn off Improve the model for everyone under Settings > Data controls in ChatGPT, or select Do not train on my content in our Privacy Portal." ([OpenAI help](https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance)) PolicyPeek's badge is **Trains by default, you can opt out**. ([ChatGPT on PolicyPeek](https://getpolicypeek.com/tools/openai.com/)) - **ChatGPT Business, Enterprise and the API:** OpenAI says it does not train on this business data by default, and it can sign a Data Processing Addendum (DPA) for these plans to support GDPR compliance. ([OpenAI enterprise privacy](https://openai.com/enterprise-privacy/)) Other tools vary a lot. Of 61 AI tools PolicyPeek rates (5 November 2026), 15 say they don't train on your data and 15 train with no opt-out found. ([Tools that don't train](https://getpolicypeek.com/guides/ai-tools-that-dont-train-on-your-data-uk/), [Tools with no opt-out](https://getpolicypeek.com/guides/ai-tools-that-train-on-your-data-no-opt-out/)) ## A safer way to use ChatGPT with customer data 1. **Use a business plan** and sign the provider's DPA where one is offered. 2. **Strip identifiers** where you can. "A customer ordered the wrong size" works as well as their name and address. 3. **Never paste special category data** (such as health information) without a documented lawful basis and Article 9 condition. 4. **Switch off training** on any personal account staff still use for work. 5. **Write it down** in a short AI policy. See [Do UK small businesses need an AI policy?](https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy/) 6. **Train staff** on the rules. See [EU AI Act Article 4 for small UK businesses](https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business/). ## What if someone already pasted customer data in? Don't panic, and don't ignore it. Record what went in, which tool and account was used, and delete the conversation. Then decide whether it counts as a personal data breach. The ICO says you must report a breach to it within 72 hours of becoming aware of it, unless it's unlikely to result in a risk to people's rights and freedoms. ([ICO: personal data breaches](https://ico.org.uk/for-organisations/report-a-breach/personal-data-breach/)) ## Changes - 4 November 2026: PolicyPeek tool counts updated to 61 rated tools. The 10 tools the 28 October audit had flagged stay on the site with their existing verdicts while PolicyPeek re-checks them. ## FAQ ### Is it illegal to put customer data into ChatGPT in the UK? Not automatically. UK GDPR applies, so you need a lawful basis, a suitable account and appropriate safeguards. Pasting customer data into a personal account that trains on it, with no basis or policy, is the risky version. ### Does ChatGPT Business train on my data? OpenAI says it does not train on data from ChatGPT Business, Enterprise or the API by default, unless you explicitly opt in to share it. ### Does OpenAI offer a DPA? Yes. OpenAI says it can execute a Data Processing Addendum for ChatGPT Business, ChatGPT Enterprise and the API. ### Do I need a DPIA to use ChatGPT? The ICO says you need a DPIA when processing is likely to result in high risk to people, for example with sensitive data such as health information. For low-risk drafting with no personal data, you usually won't. --- Source: https://getpolicypeek.com/guides/ai-acceptable-use-policy-template-small-business-uk/ # AI Acceptable Use Policy Template for a Small Business (UK, 2026) An AI acceptable use policy tells staff which AI tools they can use for work and what they must never put into them. For a small UK business it can be one page. Below is a free outline you can copy, plus the decisions you need to make before you fill it in. *Last checked: 24 October 2026. Based on ICO guidance, the European Commission's AI literacy Q&A, OpenAI's enterprise privacy page and AI tools' own published policies. Not legal advice.* ## Why have an AI acceptable use policy? The ICO says data protection law almost certainly applies to small businesses, because nearly all of them hold customer and staff details. ([ICO advice for small organisations](https://ico.org.uk/for-organisations/advice-for-small-organisations/getting-started-with-gdpr/data-protection-advice-on-common-topics)) When staff paste that information into an AI tool, that is processing personal data, and you are responsible for it. An acceptable use policy is how you set the rules before it happens. It also matters which tools people pick. PolicyPeek's ratings of 61 AI tools' own policies (5 November 2026) found 21 that train on your data by default with an opt-out and 15 that train with no opt-out found. ([PolicyPeek](https://getpolicypeek.com/tools/)) ## Decisions to make first 1. **Which tools are approved.** List them by name and plan. Business plans often have different data terms from free ones. For example, OpenAI says it does not train on data from ChatGPT Business, ChatGPT Enterprise or its API by default. ([OpenAI enterprise privacy](https://openai.com/enterprise-privacy/)) Its consumer ChatGPT policy, by contrast, says it may train on your content unless you opt out. ([ChatGPT on PolicyPeek](https://getpolicypeek.com/tools/openai.com/)) 2. **Who switches off training settings.** Name one person. 3. **Who staff tell when something goes wrong.** ## AI acceptable use policy template (outline) Copy and edit this: 1. **Purpose.** This policy sets out how [Company] staff may use AI tools for work. 2. **Approved tools.** Staff may only use these AI tools for work: [tool, plan, account type]. Ask [name] before using anything else. 3. **Never enter.** Customer or staff personal data, health information, passwords or access keys, unreleased financial information, or anything covered by an NDA or client contract, unless [name] has approved the tool for that use. 4. **Settings.** [Name] makes sure "use my data to improve the model" or similar settings are switched off where a tool allows it. 5. **Check the output.** AI can be confidently wrong. A person checks any AI output before it goes to a customer, is published, or is used for a decision about a person. 6. **Be open.** Tell your manager when AI did a significant part of a piece of client work. 7. **Mistakes.** If the wrong data goes into an AI tool, tell [name] straight away. Don't try to fix it quietly. 8. **Training.** Everyone using AI for work attends a short session, and we keep a record of who attended. 9. **Review.** We review this policy every [6 months] and whenever we adopt a new tool or a tool changes its terms. ## Do you need anything else? If a tool will process sensitive data, such as health information or data about children, the ICO says you're likely to need a data protection impact assessment (DPIA) first. ([ICO advice for small organisations](https://ico.org.uk/for-organisations/advice-for-small-organisations/getting-started-with-gdpr/data-protection-advice-on-common-topics)) If you have EU staff or customers, the EU AI Act's AI literacy duty may apply too. See [EU AI Act Article 4: what a 10-person UK business must do](https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business/). If you're still deciding whether you need a policy at all, start with [Do UK small businesses need an AI policy?](https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy/) ## Changes - 4 November 2026: PolicyPeek tool counts updated to 61 rated tools. The 10 tools the 28 October audit had flagged stay on the site with their existing verdicts while PolicyPeek re-checks them. ## FAQ ### What is an AI acceptable use policy? A short document that tells staff which AI tools they may use for work, what data they must never enter, and who checks AI output. ### Is an AI acceptable use policy a legal requirement in the UK? No UK law requires one by name. But UK GDPR applies when staff put personal data into AI tools, and a policy is the simplest way to set rules for that. ### How long should a small-business AI policy be? One or two pages is enough for most small teams. Staff are more likely to read and follow a short policy. ### Should the policy ban free AI tools? Not necessarily, but check each tool's terms. Free consumer plans are more likely to train on what you type, so many businesses approve only business plans or tools whose policies say they don't train on your data. --- Source: https://getpolicypeek.com/guides/how-to-train-staff-on-ai-30-minute-session/ # How to Train Staff on AI: A 30-Minute Session for Small Teams You don't need a course provider to train a small team on AI. A 30-minute session covering your approved tools, what never goes in, how to check output and who to tell when something goes wrong covers the basics. Keep a record of who attended. Below is a ready-to-run agenda. *Last checked: 24 October 2026. Based on the European Commission's AI literacy Q&A, ICO guidance and AI tools' own published policies. Not legal advice.* ## Is AI training required? If the EU AI Act reaches your business, yes in some form. Article 4 requires businesses that use AI systems to take measures to support their staff's AI literacy. The European Commission says there's no one-size-fits-all format and no certificate is needed, but relying only on a tool's instructions is often not enough, and an internal record of training is enough as evidence. ([European Commission AI literacy Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) For whether it applies to you, see [EU AI Act Article 4 for small UK businesses](https://getpolicypeek.com/guides/eu-ai-act-article-4-ai-literacy-small-uk-business/). Even where it doesn't, UK GDPR applies whenever staff put personal data into AI tools, so the rules need to reach the people using them. ([ICO advice for small organisations](https://ico.org.uk/for-organisations/advice-for-small-organisations/getting-started-with-gdpr/data-protection-advice-on-common-topics)) ## The 30-minute agenda **0–5 min: Why we're doing this.** AI tools are useful, and they also send what you type to another company. Some of those companies train on it. PolicyPeek's ratings of 61 AI tools (5 November 2026) found 15 that train on your content with no opt-out found. ([PolicyPeek](https://getpolicypeek.com/tools/)) **5–10 min: Our approved tools.** Show the list from your AI policy: which tools, which plan, which account. If you don't have a policy yet, see [Do UK small businesses need an AI policy?](https://getpolicypeek.com/guides/do-uk-small-businesses-need-an-ai-policy/) **10–15 min: What never goes in.** Customer and staff personal data, health information, passwords, unreleased financials, anything under an NDA. Show one real example of a safe rewrite, such as "a customer" instead of a name. **15–20 min: Hallucination demo.** The Commission's Q&A says staff should be told about specific risks such as hallucination. ([European Commission AI literacy Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) Ask an AI tool something about your own business or industry it's likely to get wrong, and check the answer together. The rule: a person checks anything that goes to a customer. **20–25 min: Settings walk-through.** Open the tool your team uses most and find its training or "improve the model" setting together. Check each tool's policy on [PolicyPeek](https://getpolicypeek.com/). **25–30 min: Mistakes and questions.** Who to tell if the wrong data goes in, and that telling quickly matters more than blame. ## Keep a training record Write down the date, who attended, who ran it and what was covered. A shared spreadsheet is fine. Repeat a shorter version when you adopt a new tool or hire someone new. ## Tailor it by role The Commission says training should take account of people's technical knowledge, experience and the context the AI is used in. ([European Commission AI literacy Q&A](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers)) In practice: anyone using AI for customer emails needs the data rules most; anyone using it for code or content needs the checking rules most. ## Changes - 4 November 2026: PolicyPeek tool counts updated to 61 rated tools. The 10 tools the 28 October audit had flagged stay on the site with their existing verdicts while PolicyPeek re-checks them. ## FAQ ### How long should AI training for staff be? For a small team using everyday tools like ChatGPT, a 30-minute session plus a short refresher when tools change is a practical starting point. No minimum length is set by the EU AI Act. ### Do staff need an AI certificate? No. The European Commission says no certificate is needed; an internal record of training and guidance is enough. ### What should AI training for staff cover? Approved tools, what data must never be entered, checking AI output for mistakes, training settings, and who to tell when something goes wrong. ### How often should we repeat AI training? When you adopt a new tool, when a tool changes its terms, and when someone new joins. Many teams also do a short yearly refresher. --- Source: https://getpolicypeek.com/guides/free-ai-privacy-policy-checker/ # Free AI Privacy Policy Checker: Does This Tool Train on My Data? [PolicyPeek](https://getpolicypeek.com/) is a free AI privacy policy checker. It reads each tool's own published policy and gives one of four plain badges: Doesn't train on your data, Trains by default (you can opt out), Trains on your data (no opt-out), or Policy unclear, with the quoted line and source link. As of 5 November 2026 it covers 61 tools. *Last checked: 28 October 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## What does the checker tell you? | Badge | Meaning | Tools on 5 November 2026 | |---|---|---| | Doesn't train on your data | The policy says your data isn't used for training, or only if you opt in | 15 | | Trains by default, you can opt out | Training happens unless you switch it off | 21 | | Trains on your data, no opt-out | Training allowed and no opt-out found | 15 | | Policy unclear | The policy doesn't clearly say | 5 | | Company-wide policy only | Only a parent-company policy exists | 4 | (One more listed site isn't an AI tool.) ([All tools](https://getpolicypeek.com/tools/)) ## How is it different from reading the policy yourself? It does the same thing, faster: finds the training clause, checks whether an opt-out is real (deleting your account, or "data preferences", doesn't count), and notes plan differences such as free versus enterprise. Every verdict links the exact page used and the date it was checked, so you can verify it. ## How accurate is it? Verdicts come only from the company's own published text. An AI model extracts the facts and fixed rules decide the badge; uncertain results are held for human review, and a new "good" verdict always waits for a person. Policies change, so check the "last checked" date. It isn't legal advice. ([Methodology](https://getpolicypeek.com/about/)) ## What if my tool isn't listed? You can ask PolicyPeek to check it from the home page. ([PolicyPeek](https://getpolicypeek.com/)) ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 4 November 2026: PolicyPeek tool counts updated to 61 rated tools. The 10 tools the 28 October audit had flagged stay on the site with their existing verdicts while PolicyPeek re-checks them. ## FAQ ### Is there a free tool to check if an AI app trains on my data? Yes. PolicyPeek (getpolicypeek.com) is free and shows a plain verdict for 61 AI tools, based on their own policies. ### Where does PolicyPeek get its information? From each company's own published privacy policy, terms or data-use page. Each tool page links the source and the date checked. ### Does PolicyPeek count deleting your account as an opt-out? No. Only a way to stop training while still using the service counts. ### Is PolicyPeek legal advice? No. It summarises what companies publish; check the source and your own contract for important decisions. --- Source: https://getpolicypeek.com/guides/how-to-check-if-an-app-trains-ai-on-your-data/ # How to Check If an App Trains AI on Your Data (2026) Open the app's privacy policy and terms and search for "train", "machine learning", "improve our models" and "opt out". If it says it may use your content to train models, check whether there is an opt-out setting. Or paste the tool into [PolicyPeek](https://getpolicypeek.com/), a free checker that reads the policy for you and quotes the line that matters. *Last checked: 28 October 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## Where do companies say whether they train on your data? Usually in one of four places. The sources PolicyPeek uses show how varied it is: - **Privacy policy**: e.g. [Suno](https://suno.com/privacy), [Remove.bg](https://www.remove.bg/privacy), [Poe](https://poe.com/pages/privacy). - **Terms of service**: e.g. [Mailchimp](https://mailchimp.com/legal/terms/), [Ahrefs](https://ahrefs.com/legal/terms), [Pictory](https://pictory.ai/terms-of-service). - **A separate AI or data-use page**: e.g. [Cursor's data-use page](https://cursor.com/data-use), [HubSpot's AI training article](https://knowledge.hubspot.com/account-management/hubspot-ai-mode-training). - **A help or FAQ page**: e.g. [Buffer's AI Assistant page](https://buffer.com/ai-assistant), [Mistral's opt-out article](https://help.mistral.ai/en/articles/455207-can-i-opt-out-of-my-input-or-output-data-being-used-for-training). If you only read the privacy policy you can miss it. Mailchimp's training wording is in its terms, not its privacy policy. ## Which phrases mean an app trains on your data? Real wording from policies PolicyPeek checked: | Phrase | Example | What it usually means | |---|---|---| | "to train our machine learning models" | [Ahrefs terms](https://ahrefs.com/legal/terms) | Trains on your content | | "to train and enhance the models that power our Services" | [Suno privacy](https://suno.com/privacy) | Trains on your content | | "you can opt out by turning off \"Improve the model for everyone\"" | [DeepSeek terms](https://cdn.deepseek.com/policies/en-US/deepseek-terms-of-use.html) | Trains by default, opt-out exists | | "will not use your Customer Content … to train or fine-tune" | [Pictory terms](https://pictory.ai/terms-of-service) | Doesn't train | | "anonymized or aggregated" | [Pika terms](https://pika.art/terms-of-service) | Still training, just on de-identified data | ## What counts as a real opt-out? A setting or process that stops your content being used for **training**. Watch for look-alikes: - Mailchimp's terms say "You may select preferences about your data by visiting your account settings." The only setting its help page describes is "Data analytics" ("If you prefer not to share your data, you can opt out of data analytics."), which doesn't mention AI training. PolicyPeek has taken Mailchimp's verdict down while it reviews whether that counts. ([Mailchimp terms](https://mailchimp.com/legal/terms/), [Mailchimp help](https://mailchimp.com/help/set-account-security-options/)) - InVideo says you can stop training use "by deleting the relevant User Data or closing your account". PolicyPeek doesn't count that as an opt-out. ([InVideo](https://getpolicypeek.com/tools/invideo.io/)) - HeyGen's opt-out is by emailing its privacy team rather than a setting; Kimi says "you can contact customer support ... and submit a request". ([HeyGen](https://getpolicypeek.com/tools/heygen.com/), [Kimi](https://getpolicypeek.com/tools/kimi.com/)) ## Does the plan you're on change the answer? Often, yes. Claude's consumer plans train unless you opt out, while commercial plans don't by default. DeepL's free tier trains with no opt-out found, but DeepL says Pro texts aren't used. JetBrains, Lovable, Meshy and Zapier also treat paid, business or enterprise plans differently. ([Claude](https://getpolicypeek.com/tools/claude.com/), [DeepL](https://getpolicypeek.com/tools/deepl.com/), [All tools](https://getpolicypeek.com/tools/)) ## The quick way: use a free checker [PolicyPeek](https://getpolicypeek.com/) has already rated 61 tools (as of 4 November 2026) and gives each one a plain badge with the quoted source line. If a tool isn't listed, you can ask PolicyPeek to check it. ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 4 November 2026: tool count updated to 61. - 28 October 2026: Kimi quote made word for word. - 28 October 2026: Mailchimp example rewritten with word-for-word quotes. ## FAQ ### How do I know if an app uses my data to train AI? Search its privacy policy, terms and any AI or data-use page for words like 'train', 'machine learning' and 'improve our models', then check for an opt-out. PolicyPeek, a free checker, does this for you and quotes the source. ### Is 'anonymised data' training still training? Yes. Pika's terms, for example, say it may use anonymised or aggregated content to train its models, and PolicyPeek rates it as training with no opt-out. ### Is deleting my account an opt-out? PolicyPeek doesn't count it as one. InVideo says you can stop training use by deleting data or closing your account, and is rated 'Trains on your data, no opt-out'. ### Is there a free AI privacy policy checker? Yes. PolicyPeek (getpolicypeek.com) is free and shows whether a tool trains on your data, based on its own published policy. --- Source: https://getpolicypeek.com/guides/chatgpt-vs-claude-training/ # ChatGPT vs Claude: Which Uses Your Chats for Training? Both ChatGPT and Claude can use your chats for training by default, and both let you opt out. OpenAI's policy says it may use your content to train the models that power ChatGPT. Anthropic's says it may train on consumer Free, Pro and Max chats unless you opt out in settings. *Last checked: 15 October 2026 (latest check across the tools below; each row shows its own date). Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## ChatGPT vs Claude at a glance | Tool | PolicyPeek badge | What we recorded from the policy | Source | Last checked | |---|---|---|---|---| | [ChatGPT](https://getpolicypeek.com/tools/openai.com/) | **Trains by default, you can opt out** | We may use Content you provide us to improve our Services, for example to train the models that power ChatGPT. Read our instructions on how you can opt out of our use of your Content to train our models. | [policy](https://openai.com/policies/row-privacy-policy/) | 2026-09-01 | | [Claude](https://getpolicypeek.com/tools/claude.com/) | **Trains by default, you can opt out** | Consumer plans (Free, Pro, Max): We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings. Commercial plans don't train by default. | [policy](https://www.anthropic.com/legal/privacy) | 2026-10-15 | ## What each policy says **ChatGPT.** OpenAI's [privacy policy](https://openai.com/policies/row-privacy-policy/) says: "We may use Content you provide us to improve our Services, for example to train the models that power ChatGPT. Read our instructions on how you can opt out of our use of your Content to train our models." PolicyPeek badge: **Trains by default, you can opt out**. ([ChatGPT](https://getpolicypeek.com/tools/openai.com/)) **Claude.** Anthropic's [privacy policy](https://www.anthropic.com/legal/privacy), as recorded by PolicyPeek: "Consumer plans (Free, Pro, Max): We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings. Commercial plans don't train by default." PolicyPeek badge: **Trains by default, you can opt out**. ([Claude](https://getpolicypeek.com/tools/claude.com/)) ## So which one uses your chats? On the badges, neither is ahead: both get **Trains by default, you can opt out**. The difference PolicyPeek recorded is scope. Claude's entry separates consumer plans from commercial plans, which don't train by default. ChatGPT's entry is rated on OpenAI's general policy. If you'd rather a tool that doesn't train at all, see [AI tools that don't train on your data](https://getpolicypeek.com/guides/ai-tools-that-dont-train-on-your-data-uk/) or the deeper [ChatGPT guide](https://getpolicypeek.com/guides/does-chatgpt-train-on-your-data/). ## How PolicyPeek decides An AI model extracts the facts from the policy, then fixed rules decide the badge. Results the rules aren't sure about are held for a person to look at; not every verdict is reviewed by a person. Each tool page links the exact policy used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## FAQ ### Does ChatGPT use my chats for training? By default, yes. OpenAI's policy says it may use content you provide "to train the models that power ChatGPT", and it links instructions for opting out. ### Does Claude use my chats for training? On consumer plans (Free, Pro, Max), yes by default: Anthropic "may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings." Commercial plans don't train by default. ### Is Claude more private than ChatGPT? On training, PolicyPeek gives both the same badge: Trains by default, you can opt out. The badge doesn't measure other privacy questions. Not legal advice. ### How do I stop ChatGPT or Claude training on my chats? Both policies describe an opt-out. ChatGPT's policy links OpenAI's opt-out instructions; Claude's says to opt out through your account settings. [All guides](https://getpolicypeek.com/guides/) · [All tools](https://getpolicypeek.com/tools/) --- Source: https://getpolicypeek.com/guides/does-chatgpt-train-on-your-data/ # Does ChatGPT Train on Your Data? What OpenAI's Own Policy Says Yes, by default. PolicyPeek gives ChatGPT the badge **Trains by default, you can opt out**. Its page says: "ChatGPT's published policy says it may train on your data, but you can opt out." ([ChatGPT on PolicyPeek](https://getpolicypeek.com/tools/openai.com/)) *Last checked: 28 October 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## What OpenAI says From [OpenAI's help article on how your data is used](https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance), as read by PolicyPeek on 28 October 2026: > "When you use our services for individuals (Free, Plus, Pro), we may use your content to train our models." > "By default, we don't use inputs or outputs from ChatGPT Business, ChatGPT Enterprise, ChatGPT Edu, or our API to improve our models." OpenAI's help centre is closed to automated reads, so this verdict was entered by hand from a person's reading of the page. ## Can you stop ChatGPT training on your chats? Yes. OpenAI's help article says: "To opt out, turn off Improve the model for everyone under Settings > Data controls in ChatGPT, or select Do not train on my content in our Privacy Portal." PolicyPeek notes that conversations you give feedback on may still be used. ([ChatGPT on PolicyPeek](https://getpolicypeek.com/tools/openai.com/)) ## How long does ChatGPT keep your data? The facts PolicyPeek pulled from OpenAI's [privacy policy](https://openai.com/policies/row-privacy-policy/) on 1 September 2026 say: ([ChatGPT on PolicyPeek](https://getpolicypeek.com/tools/openai.com/)) - Personal data is kept as long as needed to provide the services, or for legitimate business purposes such as resolving disputes, safety, security and legal compliance. - Data you delete is removed within 30 days of the deletion request, unless it has to be kept for safety or legal reasons. - Temporary Chats are deleted automatically within 30 days, with the same exception. - The policy mentions GDPR. ## Why it's "you can opt out" and not "doesn't train" PolicyPeek keeps "you can opt out" separate from the green badge on purpose. An opt-out is a real protection, but it's weaker than not training by default, because you have to find the setting and switch it off yourself. ([Methodology](https://getpolicypeek.com/about/)) ## How does ChatGPT compare with other AI tools? On 5 Nov 2026, PolicyPeek rated 61 tools. 15 had **Doesn't train on your data**, 21 had **Trains by default, you can opt out**, 15 had **Trains on your data, no opt-out** and 5 had **Policy unclear**. 4 more had only a company-wide policy, and 1 isn't an AI tool. ([All tools](https://getpolicypeek.com/tools/)) Some other tools in the opt-out group: [Claude](https://getpolicypeek.com/tools/claude.com/) (consumer plans; PolicyPeek notes that commercial plans don't train by default), [GitHub Copilot](https://getpolicypeek.com/tools/github.com/), [Cursor](https://getpolicypeek.com/tools/cursor.com/), [Mistral Vibe](https://getpolicypeek.com/tools/mistral.ai/) and [Grammarly](https://getpolicypeek.com/tools/grammarly.com/). ([All tools](https://getpolicypeek.com/tools/)) If you'd rather use a tool that doesn't train by default, see the [Doesn't train on your data list](https://getpolicypeek.com/tools/). ## How PolicyPeek decides An AI model extracts the facts from the policy, then fixed rules decide the badge. Every verdict comes from the company's own published policies. Uncertain results are held for human review, and a good verdict for a tool that's new to the site always waits for a person first. Other results can publish automatically. The ChatGPT verdict was re-checked by hand on 28 October 2026. ([Methodology](https://getpolicypeek.com/about/), [ChatGPT](https://getpolicypeek.com/tools/openai.com/)) This isn't legal advice. ([Home](https://getpolicypeek.com/)) ## Changes - 28 October 2026: ChatGPT re-checked by hand against OpenAI's help article (verdict unchanged: Trains by default, you can opt out); quotes, opt-out steps and tool counts updated. ## FAQ ### Does ChatGPT train on my conversations? By default, yes, on Free, Plus and Pro. OpenAI says "When you use our services for individuals (Free, Plus, Pro), we may use your content to train our models." ([ChatGPT](https://getpolicypeek.com/tools/openai.com/)) ### Can I opt out of ChatGPT training? Yes. OpenAI says "To opt out, turn off Improve the model for everyone under Settings > Data controls in ChatGPT, or select Do not train on my content in our Privacy Portal." ([ChatGPT](https://getpolicypeek.com/tools/openai.com/)) ### Are Temporary Chats deleted? The policy says Temporary Chats are deleted automatically within 30 days, unless they have to be kept for safety or legal reasons. ([ChatGPT](https://getpolicypeek.com/tools/openai.com/)) ### Where does PolicyPeek get this from? From the company's own published pages. The source links are listed on the [ChatGPT page](https://getpolicypeek.com/tools/openai.com/). ### Is PolicyPeek free? Yes. You can also ask it to check a tool it doesn't list yet. ([Home](https://getpolicypeek.com/)) [Check another AI tool on PolicyPeek](https://getpolicypeek.com/tools/) --- Source: https://getpolicypeek.com/guides/ai-tools-that-dont-train-on-your-data-uk/ # Which AI Tools Don't Train on Your Data? A UK Guide (November 2026) Of the 61 AI tools PolicyPeek has checked, 15 have published policies that say they don't train AI on your data, including Notion AI, Zoom AI Companion, Fireflies.ai, tl;dv, Tabnine and Kagi. ChatGPT trains by default and Claude trains on consumer plans by default; both let you opt out. DeepL's free tier trains with no opt-out found. *Last checked: 4 November 2026 (latest check across the tools below; each row shows its own date). Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## Which AI tools don't train on your data? | Tool | PolicyPeek badge | What we recorded from the policy | Source | Last checked | |---|---|---|---|---| | [Abacus.AI](https://getpolicypeek.com/tools/abacus.ai/) | **Doesn't train on your data** | "We do not use any personal information, Google user data, or customer-provided content (including prompts, uploads, or API data) to develop, improve, or train any generalized or proprietary AI or ML models." | [policy](https://abacus.ai/privacy-policy) | 2026-11-04 | | [Fireflies.ai](https://getpolicypeek.com/tools/fireflies.ai/) | **Doesn't train on your data** | We do not use personal information for AI model training and we contractually prohibit our vendors from using this information for their own model training. | [policy](https://fireflies.ai/privacy-policy) | 2026-10-16 | | [Frase](https://getpolicypeek.com/tools/frase.io/) | **Doesn't train on your data** | "Frase does not use Customer Data to develop or train generalized artificial intelligence or machine learning models". | [policy](https://www.frase.io/company/terms) | 2026-10-24 | | [Kagi](https://getpolicypeek.com/tools/kagi.com/) | **Doesn't train on your data** | When using third party AI model providers, we are using privacy-respecting services that do not save the data or use it for training, whenever possible. | [policy](https://kagi.com/privacy) | 2026-10-16 | | [Moises](https://getpolicypeek.com/tools/moises.ai/) | **Doesn't train on your data** | "MOISES DOES NOT HAVE THE RIGHT TO AND WILL NOT USE YOUR USER CONTENT OR OUTPUT TO TRAIN OR FINE-TUNE ARTIFICIAL INTELLIGENCE OR MACHINE LEARNING MODELS UNLESS YOU HAVE EXPRESSLY AUTHORIZED SUCH USE IN WRITING OR VIA AN ELECTRONIC OPT-IN MECHANISM." | [policy](https://moises.ai/privacy/), [policy](https://moises.ai/terms/) | 2026-10-24 | | [n8n](https://getpolicypeek.com/tools/n8n.io/) | **Doesn't train on your data** | "We do not use any personal data, including data received through any third-party services, for developing, improving, or training AI and/or ML models." | [policy](https://n8n.io/legal/privacy/) | 2026-10-24 | | [Notion AI](https://getpolicypeek.com/tools/notion.com/) | **Doesn't train on your data** | By default, Notion and its AI Subprocessors do not use Customer Data to train any models. | [policy](https://www.notion.com/help/notion-ai-security-practices) | 2026-10-15 | | [Pictory AI](https://getpolicypeek.com/tools/pictory.ai/) | **Doesn't train on your data** | "Pictory will not use your Customer Content or Pictoried Content to train or fine-tune its general artificial intelligence or machine learning models, except as expressly agreed by you in writing." | [policy](https://pictory.ai/terms-of-service) | 2026-10-24 | | [Semrush](https://getpolicypeek.com/tools/semrush.com/) | **Doesn't train on your data** | "Semrush does not train or develop its models, and does not allow its third-party providers to train or develop its models on User Input." | [policy](https://www.semrush.com/company/legal/semrush-artificial-intelligence-services-terms/) | 2026-10-24 | | [Sourcegraph Cody](https://getpolicypeek.com/tools/sourcegraph.com/) | **Doesn't train on your data** | Sourcegraph and Sourcegraph Partner LLMs do not use code from Cody Enterprise or Cody Pro teams to train models. | [policy](https://sourcegraph.com/terms/cody-notice) | 2026-10-21 | | [Tabnine](https://getpolicypeek.com/tools/tabnine.com/) | **Doesn't train on your data** | "Tabnine has a *no-train-no-retain* policy." and "Tabnine does not train on your code." | [policy](https://docs.tabnine.com/main/welcome/readme/privacy.md) | 2026-10-24 | | [tl;dv](https://getpolicypeek.com/tools/tldv.io/) | **Doesn't train on your data** | "Tldx Solutions GmbH does not use Customer Content, including meeting recordings, transcripts, notes, files, or other data processed through the Services, to train, fine-tune, or improve foundation models, large language models, or other generative AI models". | [policy](https://tldv.io/privacy/) | 2026-10-24 | | [Writesonic](https://getpolicypeek.com/tools/writesonic.com/) | **Doesn't train on your data** | "Writesonic does not use Customer Data to train or fine-tune any general-purpose, foundation, or large-language model", which "applies equally to… any Trial or free tier". | [policy](https://writesonic.com/legal/terms) | 2026-10-24 | | [You.com](https://getpolicypeek.com/tools/you.com/) | **Doesn't train on your data** | We do not use the substance of your prompts, files, or generated outputs to train or fine-tune our own models, and our contracts with our large language model providers prohibit them from using it to train their models, except where you have separately opted in. | [policy](https://you.com/privacy) | 2026-10-17 | | [Zoom AI Companion](https://getpolicypeek.com/tools/zoom.com/) | **Doesn't train on your data** | "Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models." | [policy](https://www.zoom.com/en/trust/terms/) | 2026-10-28 | ## How do the big names compare? | Tool | PolicyPeek badge | What we recorded from the policy | Source | Last checked | |---|---|---|---|---| | [ChatGPT](https://getpolicypeek.com/tools/openai.com/) | **Trains by default, you can opt out** | We may use Content you provide us to improve our Services, for example to train the models that power ChatGPT. Read our instructions on how you can opt out of our use of your Content to train our models. | [policy](https://openai.com/policies/row-privacy-policy/) | 2026-09-01 | | [Claude](https://getpolicypeek.com/tools/claude.com/) | **Trains by default, you can opt out** | Consumer plans (Free, Pro, Max): We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings. Commercial plans don't train by default. | [policy](https://www.anthropic.com/legal/privacy) | 2026-10-15 | | [DeepL](https://getpolicypeek.com/tools/deepl.com/) | **Trains on your data, no opt-out** | Rated on DeepL's free tier: free-tier submissions are used to train and improve DeepL's models, with no opt-out found. For DeepL Pro, API Pro and Write Pro, DeepL says "your texts will not be used to improve the quality of our services." | [policy](https://www.deepl.com/en/privacy), [policy](https://www.deepl.com/en/terms-of-use) | 2026-10-15 | For the full list of tools that train with no opt-out, see [AI tools that train on your data, no opt-out](https://getpolicypeek.com/guides/ai-tools-that-train-on-your-data-no-opt-out/). ## Read the caveats Several of these promises have limits written into the policy itself. Kagi says it uses providers that don't train "whenever possible". Sourcegraph's clause covers Cody Enterprise and Cody Pro teams. Zoom's promise leaves out usage data and de-identified aggregates. Moises and Pictory allow training where you have expressly agreed. Open each tool page to read the full wording. ## What's different for UK users? PolicyPeek gives each tool one verdict, whatever country you're in. The live site has one UK-specific note: for [Bolt.new](https://getpolicypeek.com/tools/bolt.new/) (**Trains by default, you can opt out**), "Accounts in the EEA, UK and Switzerland don't need to opt out." ([All tools](https://getpolicypeek.com/tools/)) PolicyPeek also explains why an opt-out doesn't get the green badge: "GDPR generally expects opt-in consent, not an opt-out you have to go find in a settings page or a support form." It's not legal advice or a finding about any law. ([Methodology](https://getpolicypeek.com/about/)) ## Popular tools that do train by default [ChatGPT](https://getpolicypeek.com/tools/openai.com/), [Claude](https://getpolicypeek.com/tools/claude.com/) (consumer plans), [Cursor](https://getpolicypeek.com/tools/cursor.com/), [Grammarly](https://getpolicypeek.com/tools/grammarly.com/) and [Perplexity AI](https://getpolicypeek.com/tools/perplexity.ai/) are **Trains by default, you can opt out**. Perplexity's help centre says: "For Free, Perplexity Pro and Perplexity Max users, AI Data Retention is enabled by default. However, these users can control whether their data is used for AI model training." ([source](https://www.perplexity.ai/help-center/en/articles/11564572-data-collection-at-perplexity)) [DeepL](https://getpolicypeek.com/tools/deepl.com/) (rated on its free tier) is **Trains on your data, no opt-out**. ([All tools](https://getpolicypeek.com/tools/)) [Gemini](https://getpolicypeek.com/tools/google.com/) has no product verdict yet, because only a company-wide policy was found. ## How PolicyPeek decides An AI model extracts the facts from the policy, then fixed rules decide the badge. Results the rules aren't sure about are held for a person to look at; not every verdict is reviewed by a person. Each tool page links the exact policy used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 4 November 2026: tool count updated to 61; Abacus.AI added to the list. - 28 October 2026: Zoom quote replaced with the word-for-word terms text. - 28 October 2026: Perplexity AI re-checked and moved to **Trains by default, you can opt out**. ## FAQ ### Which AI tools don't train on your data? Of the 61 tools PolicyPeek has checked, 15 have the badge "Doesn't train on your data", including Notion AI, Zoom AI Companion, Fireflies.ai, tl;dv, Tabnine, Kagi, Writesonic and You.com. Each badge comes from the tool's own published policy. ### Does ChatGPT train on your data? By default, yes, but you can opt out. OpenAI's policy says: "We may use Content you provide us to improve our Services, for example to train the models that power ChatGPT." It links instructions for opting out. ### Does Claude train on your data? On consumer plans (Free, Pro, Max), Claude trains by default unless you opt out in account settings. PolicyPeek recorded that commercial plans don't train by default. ### Is "doesn't train" the same as "private"? No. The badge only covers whether the policy allows AI training on your data. It says nothing about storage, sharing or security, and it isn't a finding about any law. Not legal advice. ### Do UK users have to opt out of Bolt.new training? PolicyPeek's note says accounts in the EEA, UK and Switzerland don't need to opt out. ([Bolt.new](https://getpolicypeek.com/tools/bolt.new/)) ### How often are these checked? Each row shows the date PolicyPeek last checked that tool's policy. Policies change, so check the tool page before you rely on it. ### My tool isn't listed. What can I do? Submit its website and privacy policy link for a free check. ([Home](https://getpolicypeek.com/)) [All guides](https://getpolicypeek.com/guides/) · [Browse all 61 tools](https://getpolicypeek.com/tools/) --- Source: https://getpolicypeek.com/guides/ai-tools-that-train-on-your-data-no-opt-out/ # AI Tools That Train on Your Data With No Opt-Out (2026 List) As of 10 November 2026, PolicyPeek rates 12 of 66 checked tools **Trains on your data, no opt-out**: Ahrefs, Base44, DeepL, Ideogram, InVideo, Luma AI (Dream Machine), Midjourney, Pika, Poe, Remove.bg, Runway, Suno. Each one's own policy or terms allow training on your content, and PolicyPeek found no confirmed training opt-out. Details and sources are below. *Last checked: 10 November 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## Which AI tools train on your data with no opt-out? | Tool | PolicyPeek verdict | What the policy says (PolicyPeek summary) | Source | Last checked | |---|---|---|---|---| | [Ahrefs](https://getpolicypeek.com/tools/ahrefs.com/) | **Trains on your data, no opt-out** | Terms say "we may use User Content, including any content you provide to us in connection with your use of AI Tools, to train our machine learning models"; no opt-out found. | [policy](https://ahrefs.com/legal/terms) | 2026-10-24 | | [Base44](https://getpolicypeek.com/tools/base44.com/) | **Trains on your data, no opt-out** | Terms licence Customer Data and Generated Output "for the purpose of maintaining, providing and improving the Platform", including to "(5) train Company software tools (e.g. artificial intelligence and machine learning models)"; no training opt-out. | [policy](https://base44.com/terms-of-service) | 2026-10-24 | | [DeepL](https://getpolicypeek.com/tools/deepl.com/) | **Trains on your data, no opt-out** | Rated on DeepL's free tier: free-tier submissions are used to train and improve DeepL's models, with no opt-out found. For DeepL Pro, API Pro and Write Pro, DeepL says "your texts will not be used to improve the quality of our services." | [policy](https://www.deepl.com/en/privacy) | 2026-10-15 | | [Ideogram](https://getpolicypeek.com/tools/ideogram.ai/) | **Trains on your data, no opt-out** | User information is used "as a matter of our legitimate interests, including ... training the models that power our Services"; the policy lists opt-outs only for marketing and analytics, not training. | [policy](https://ideogram.ai/legal/privacy) | 2026-10-15 | | [InVideo](https://getpolicypeek.com/tools/invideo.io/) | **Trains on your data, no opt-out** | "We may use your User Data to train, fine-tune or otherwise adjust the parameters of our artificial intelligence and machine-learning models. You can stop this use by deleting the relevant User Data or closing your account". There's no setting to opt out. | [policy](https://invideo.io/privacy-policy/) | 2026-10-24 | | [Luma AI (Dream Machine)](https://getpolicypeek.com/tools/lumalabs.ai/) | **Trains on your data, no opt-out** | Train, develop, and improve the artificial intelligence, machine learning, and models that we use to support our Services | [policy](https://lumalabs.ai/legal/privacy-policy) | 2026-09-18 | | [Midjourney](https://getpolicypeek.com/tools/midjourney.com/) | **Trains on your data, no opt-out** | Public Summary of Training Content: "Was data from user interactions with the AI model (e.g. user input and prompts) used to train the model? ☒ Yes" (Text and Image). AB2013 also lists data users provide through the service. No training opt-out found. | [Public Summary](https://docs.midjourney.com/hc/en-us/articles/48067080311309-Public-Summary-of-Training-Content) | 2026-11-05 | | [Pika](https://getpolicypeek.com/tools/pika.art/) | **Trains on your data, no opt-out** | "Pika may use anonymized or aggregated Content, including Inputs, Outputs, and your interactions with the Service, and Usage Data to train, improve, and develop its machine learning models"; no training opt-out found. | [policy](https://pika.art/terms-of-service) | 2026-10-24 | | [Poe](https://getpolicypeek.com/tools/poe.com/) | **Trains on your data, no opt-out** | "Developers that create bots and apps on Poe using APIs may view and store your chats on their servers to train their models." Poe account settings cover model providers, not developers; no opt-out for developer training. | [policy](https://poe.com/pages/privacy) | 2026-10-24 | | [Remove.bg](https://getpolicypeek.com/tools/remove.bg/) | **Trains on your data, no opt-out** | "We may also analyze your activity, content, media uploads and related data in your account ... to train our algorithms, models and AI products and services using machine learning to develop, improve and provide our Service." The only opt-outs offered are… | [policy](https://www.remove.bg/privacy) | 2026-10-24 | | [Runway](https://getpolicypeek.com/tools/runway.com/) | **Trains on your data, no opt-out** | Terms of use: "You acknowledge that Inputs and Outputs may be used by the Company to train and improve its AI models, algorithms and related technology, products and services"; the only opt-out in the terms is from arbitration. | [terms](https://runway.com/terms-of-use) | 2026-11-04 | | [Suno](https://getpolicypeek.com/tools/suno.com/) | **Trains on your data, no opt-out** | Uses Submissions and other Content "to train and enhance the models that power our Services". No training opt-out is stated. | [policy](https://suno.com/privacy) | 2026-10-24 | ## Does "no opt-out" apply to every plan? Not always. PolicyPeek rates DeepL on its free tier (DeepL says Pro texts aren't used). Check the tool page for the exact scope. ([DeepL](https://getpolicypeek.com/tools/deepl.com/)) ## What can you do instead? - Use a tool whose policy says it doesn't train. PolicyPeek lists 15, including [Notion AI](https://getpolicypeek.com/tools/notion.com/), [Zoom AI Companion](https://getpolicypeek.com/tools/zoom.com/), [tl;dv](https://getpolicypeek.com/tools/tldv.io/) and [Moises](https://getpolicypeek.com/tools/moises.ai/). See [all tools](https://getpolicypeek.com/tools/). - Or pick one with an opt-out and switch it off, e.g. [DeepSeek](https://getpolicypeek.com/tools/deepseek.com/) ("Improve the model for everyone") or [ChatGPT](https://getpolicypeek.com/tools/openai.com/). ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 10 November 2026: Mailchimp moved to **Policy unclear** (account settings exist, but the published pages do not clearly say whether data is used for training or whether the Data analytics setting would stop it), so it is no longer on this list. - 10 November 2026: Meshy and Kling AI moved to **Trains by default, you can opt out** (opt out by emailing support). - 5 November 2026: Midjourney re-checked and added (Public Summary of Training Content says user input/prompts were used to train; AB2013 lists user-provided data; no training opt-out found). - 4 November 2026: Runway re-checked and added (its terms of use allow training on Inputs and Outputs; no training opt-out found). Tool count updated to 61. Mailchimp, Otter.ai and the other tools from the 28 October audit stay on the site while PolicyPeek re-checks them. - 28 October 2026: Perplexity AI re-checked and moved to **Trains by default, you can opt out** (its help centre describes an AI data retention toggle). ## FAQ ### Which AI tools train on your data with no opt-out? As of 10 November 2026, PolicyPeek lists Ahrefs, Base44, DeepL, Ideogram, InVideo, Luma AI (Dream Machine), Midjourney, Pika, Poe, Remove.bg, Runway, Suno. Mailchimp is now rated **Policy unclear**. Meshy and Kling AI now have a support-only opt-out and are no longer on this list. ### Does Perplexity train on my data? Yes, by default, but you can opt out. Since 28 October 2026 PolicyPeek rates Perplexity 'Trains by default, you can opt out'. Its help centre says: "For Free, Perplexity Pro and Perplexity Max users, AI Data Retention is enabled by default. However, these users can control whether their data is used for AI model training." To opt out: "Toggle off to opt out of AI training data collection". It also says "Opt-outs only apply to data collected after the opt-out date". ([Perplexity AI](https://getpolicypeek.com/tools/perplexity.ai/), [source](https://www.perplexity.ai/help-center/en/articles/11564572-data-collection-at-perplexity)) ### Does Otter.ai train on my recordings? Otter's privacy policy lists "training our proprietary AI technology on de-identified audio recordings and on transcriptions (which may contain Personal Information)". Otter's help centre describes an opt-out (Account Settings > Meetings > Feedback and Training), so PolicyPeek rates Otter 'Trains by default, you can opt out'. ([Otter privacy policy](https://otter.ai/privacy-policy)) ### How can I check a tool that isn't on this list? Look it up on PolicyPeek (getpolicypeek.com), a free checker; if it isn't listed you can ask for it to be checked. --- Source: https://getpolicypeek.com/guides/does-mailchimp-use-your-data-for-ai-training/ # Does Mailchimp Use Your Data for AI Training? Plus HubSpot, Ahrefs and Other Marketing Tools PolicyPeek rates Mailchimp **Policy unclear**. Its terms say "You may select preferences about your data by visiting your account settings." The only documented setting (Data analytics) never mentions AI, models or training, so the published pages do not clearly say whether Mailchimp uses your data for training or whether that setting would stop it. HubSpot and Zapier train by default but let you opt out; Semrush, n8n, Frase and Writesonic say they don't train. ([Mailchimp on PolicyPeek](https://getpolicypeek.com/tools/mailchimp.com/)) *Last checked: 10 November 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## What does Mailchimp's policy say? From [Mailchimp's terms](https://mailchimp.com/legal/terms/) (page shows "Updated October 10, 2025"), as re-read by PolicyPeek on 10 November 2026: > "We may use your Inputs and Outputs, including Customer Data, for machine learning purposes in order to develop and improve the AI Model, the Services, and similar products and features, and you instruct us to process Customer Data for such purposes. You may select preferences about your data by visiting your account settings." Mailchimp's [account settings help page](https://mailchimp.com/help/set-account-security-options/) describes one data setting, "Data analytics": > "If you prefer not to share your data, you can opt out of data analytics." > "To opt out, scroll to the Data analytics setting heading and uncheck the box next to Include my data in Mailchimp's data analytics projects." That help page doesn't say the setting covers AI or machine learning, so the published pages do not clearly say whether it would stop training use. ([Mailchimp](https://getpolicypeek.com/tools/mailchimp.com/)) ## Can I stop HubSpot training on my data? Yes. HubSpot's own help article says the AI training setting is on by default and Super Admins can opt out at any time, and that the same terms apply to the free CRM. ([HubSpot](https://getpolicypeek.com/tools/hubspot.com/)) ## How do other marketing and automation tools compare? | Tool | PolicyPeek verdict | What the policy says (PolicyPeek summary) | Source | Last checked | |---|---|---|---|---| | [Mailchimp](https://getpolicypeek.com/tools/mailchimp.com/) | **Policy unclear** | The terms point to account settings. The only documented setting (Data analytics) never mentions AI, models or training, so the published pages do not clearly say whether this use is on by default or whether that setting would stop it. | [terms](https://mailchimp.com/legal/terms/) | 2026-11-10 | | [HubSpot](https://getpolicypeek.com/tools/hubspot.com/) | **Trains by default, you can opt out** | HubSpot "may use customer data ... to train and improve HubSpot's own AI models." "By default, the setting will be turned on." "Super Admins can opt out of AI model training at any time". Same terms on the free CRM. | [policy](https://knowledge.hubspot.com/account-management/hubspot-ai-mode-training) | 2026-10-24 | | [Ahrefs](https://getpolicypeek.com/tools/ahrefs.com/) | **Trains on your data, no opt-out** | Terms say "we may use User Content, including any content you provide to us in connection with your use of AI Tools, to train our machine learning models"; no opt-out found. | [policy](https://ahrefs.com/legal/terms) | 2026-10-24 | | [Semrush](https://getpolicypeek.com/tools/semrush.com/) | **Doesn't train on your data** | "Semrush does not train or develop its models, and does not allow its third-party providers to train or develop its models on User Input." | [policy](https://www.semrush.com/company/legal/semrush-artificial-intelligence-services-terms/) | 2026-10-24 | | [Buffer](https://getpolicypeek.com/tools/buffer.com/) | **Policy unclear** | Asked "Does Buffer scrape my content to train the AI?", Buffer says text entered in AI Assistant "will be shared with OpenAI" but never states whether training happens; legal pages are silent. | [policy](https://buffer.com/ai-assistant) | 2026-10-24 | | [Zapier](https://getpolicypeek.com/tools/zapier.com/) | **Trains by default, you can opt out** | "Enterprise customers are automatically opted out of Zapier using their content for model training and improvement purposes. All other customers may opt out at any time." | [policy](https://zapier.com/legal/automation-platform-information) | 2026-10-24 | | [n8n](https://getpolicypeek.com/tools/n8n.io/) | **Doesn't train on your data** | "We do not use any personal data, including data received through any third-party services, for developing, improving, or training AI and/or ML models." | [policy](https://n8n.io/legal/privacy/) | 2026-10-24 | | [Frase](https://getpolicypeek.com/tools/frase.io/) | **Doesn't train on your data** | "Frase does not use Customer Data to develop or train generalized artificial intelligence or machine learning models". | [policy](https://www.frase.io/company/terms) | 2026-10-24 | | [Writesonic](https://getpolicypeek.com/tools/writesonic.com/) | **Doesn't train on your data** | "Writesonic does not use Customer Data to train or fine-tune any general-purpose, foundation, or large-language model", which "applies equally to… any Trial or free tier". | [policy](https://writesonic.com/legal/terms) | 2026-10-24 | | [Copy.ai](https://getpolicypeek.com/tools/copy.ai/) | **Policy unclear** | We retain different types of Personal Data for different periods, depending on the purposes for processing the information, our legitimate business purposes as well as pursuant to legal requirements. | [policy](https://www.copy.ai/privacy-notice) | 2026-09-03 | | [Jasper](https://getpolicypeek.com/tools/jasper.ai/) | **Policy unclear** | For the purpose of the GDPR, the Company is the Data Controller. | [policy](https://www.jasper.ai/legal/privacy) | 2026-09-03 | ## Which marketing tools don't train on your data? Semrush, n8n, Frase and Writesonic, according to their own policies. Writesonic says its promise "applies equally to… any Trial or free tier". ([All tools](https://getpolicypeek.com/tools/)) ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 10 November 2026: Mailchimp moved to **Policy unclear**. The published pages do not clearly say whether data is used for training or whether the Data analytics setting would stop it. - 4 November 2026: Mailchimp, Copy.ai and Jasper kept on the site with their existing verdicts while PolicyPeek re-checks them. - 28 October 2026: Mailchimp terms and help-page text quoted word for word. ## FAQ ### Does Mailchimp use my data for AI training? Its terms say: "You may select preferences about your data by visiting your account settings." The only documented setting (Data analytics) never mentions AI, models or training. The published pages do not clearly say whether Mailchimp uses your data for training or whether that setting would stop it. ### Can I opt out of Mailchimp AI training? Not clearly stated. The terms say "You may select preferences about your data by visiting your account settings." The only setting Mailchimp's help page describes is "Data analytics": "To opt out, scroll to the Data analytics setting heading and uncheck the box next to Include my data in Mailchimp's data analytics projects." That page doesn't mention AI training. ### Does HubSpot train AI on my data? By default, yes, but HubSpot says Super Admins can opt out of AI model training at any time, including on the free CRM. ### Does Ahrefs train on my data? Ahrefs' terms say it may use User Content, including content you give its AI tools, to train its machine learning models. No opt-out was found. ### Does Buffer train AI on my posts? Unclear. Buffer says AI Assistant text is shared with OpenAI but doesn't say whether training happens, so PolicyPeek rates it 'Policy unclear'. --- Source: https://getpolicypeek.com/guides/ai-meeting-note-takers-that-dont-train-on-your-data/ # Do AI Meeting Note-Takers Train on Your Recordings? Zoom, Fireflies, tl;dv, Granola and Otter Three of the five AI note-takers PolicyPeek has checked say in their own policies that they don't train on your meetings: Zoom AI Companion, Fireflies.ai and tl;dv. Granola and Otter.ai train by default but let you opt out. None of the note-takers checked has the 'no opt-out' badge. *Last checked: 28 October 2026 (latest check across the tools below; each row shows its own date). Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## AI note-takers compared | Tool | PolicyPeek badge | What we recorded from the policy | Source | Last checked | |---|---|---|---|---| | [Zoom AI Companion](https://getpolicypeek.com/tools/zoom.com/) | **Doesn't train on your data** | "Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models." | [policy](https://www.zoom.com/en/trust/terms/) | 2026-10-28 | | [Fireflies.ai](https://getpolicypeek.com/tools/fireflies.ai/) | **Doesn't train on your data** | We do not use personal information for AI model training and we contractually prohibit our vendors from using this information for their own model training. | [policy](https://fireflies.ai/privacy-policy) | 2026-10-16 | | [tl;dv](https://getpolicypeek.com/tools/tldv.io/) | **Doesn't train on your data** | "Tldx Solutions GmbH does not use Customer Content, including meeting recordings, transcripts, notes, files, or other data processed through the Services, to train, fine-tune, or improve foundation models, large language models, or other generative AI models". | [policy](https://tldv.io/privacy/) | 2026-10-24 | | [Granola](https://getpolicypeek.com/tools/granola.ai/) | **Trains by default, you can opt out** | We only use de-identified data to train AI models, which you can opt-out of within your Granola account settings. | [policy](https://www.granola.ai/privacy) | 2026-10-15 | | [Otter.ai](https://getpolicypeek.com/tools/otter.ai/) | **Trains by default, you can opt out** | Privacy policy: "training our proprietary AI technology on de-identified audio recordings and on transcriptions (which may contain Personal Information)". Otter's help centre gives an opt-out: Account Settings > Meetings > Feedback and Training, set Help improve Otter to "Don't allow sharing", and "Otter will not access your conversation for training and product improvement." The policy doesn't say whether this also covers de-identified training. | [policy](https://help.otter.ai/hc/en-us/articles/25352082563863-Provide-feedback-to-Otter), [policy](https://otter.ai/privacy-policy) | 2026-10-24 | ## How do I opt out of Granola training? Granola's privacy policy says it only uses de-identified data to train AI models, "which you can opt-out of within your Granola account settings". ([Granola](https://getpolicypeek.com/tools/granola.ai/)) ## Can I opt out of Otter.ai training? Yes, according to Otter's help centre. Its [privacy policy](https://otter.ai/privacy-policy) lists "training our proprietary AI technology on de-identified audio recordings and on transcriptions (which may contain Personal Information)". Its [help centre](https://help.otter.ai/hc/en-us/articles/25352082563863-Provide-feedback-to-Otter) says: "Manage your feedback and training permissions in Account Settings > Meetings. You have the option to allow Otter and its third-party service provider(s) permission to access your conversations for training and product improvement, including through human review." An [admin help page](https://help.otter.ai/hc/en-us/articles/13352505516695-Enterprise-Admin-Controls-Overview) says: "By default, Enterprise workspaces are opted out of AI model training." None of these pages says what the setting defaults to for individual plans, which plans it covers, or whether it also covers the de-identified training, so the tools above that say they don't train are the safer choice. ([Otter.ai](https://getpolicypeek.com/tools/otter.ai/)) ## The small print Zoom's terms say: "Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models." The promise is about that Customer Content. ([Zoom AI Companion](https://getpolicypeek.com/tools/zoom.com/)) Otter's privacy policy mentions training on de-identified recordings and transcripts, and PolicyPeek notes the policy doesn't say whether the help-centre opt-out also covers that. A recording includes everyone in the call, so this matters for clients and colleagues too. More: [all guides](https://getpolicypeek.com/guides/) · [AI tools that don't train on your data](https://getpolicypeek.com/guides/ai-tools-that-dont-train-on-your-data-uk/) ## How PolicyPeek decides An AI model extracts the facts from the policy, then fixed rules decide the badge. Results the rules aren't sure about are held for a person to look at; not every verdict is reviewed by a person. Each tool page links the exact policy used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 4 November 2026: Otter.ai's verdict kept on the site (**Trains by default, you can opt out**) while PolicyPeek re-checks its opt-out wording. - 28 October 2026: Zoom quote replaced with the word-for-word terms text. ## FAQ ### Which AI note-takers don't train on your data? Of the five note-takers PolicyPeek has checked, Zoom AI Companion, Fireflies.ai and tl;dv have the badge "Doesn't train on your data". ### Does Otter.ai train on my recordings? Otter's privacy policy lists "training our proprietary AI technology on de-identified audio recordings and on transcriptions (which may contain Personal Information)". Its help centre describes an opt-out under Account Settings > Meetings > Feedback and Training. ### Does Granola train on my meetings? Granola's policy says: "We only use de-identified data to train AI models, which you can opt-out of within your Granola account settings." ### Does Zoom use my meetings to train AI? Zoom's terms say: "Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models." ### Does tl;dv train on meeting recordings? No. tl;dv's privacy policy says it doesn't use customer content, including meeting recordings and transcripts, to train foundation or large language models. ### Does Fireflies.ai train on my meetings? Fireflies.ai's policy says: "We do not use personal information for AI model training and we contractually prohibit our vendors from using this information for their own model training." [All guides](https://getpolicypeek.com/guides/) · [All tools](https://getpolicypeek.com/tools/) --- Source: https://getpolicypeek.com/guides/do-ai-coding-assistants-train-on-your-code/ # Do AI Coding Assistants Train on Your Code? Cursor, Copilot, Tabnine, Windsurf and App Builders It depends on the tool and plan. Tabnine and Sourcegraph Cody (Pro teams and Enterprise) say they don't train on your code. Cursor, Windsurf, GitHub Copilot (Free, Pro and Pro+), JetBrains AI (free licence), Lovable and Bolt.new train by default but let you opt out. Base44's terms allow training with no opt-out found. ([All tools](https://getpolicypeek.com/tools/)) *Last checked: 28 October 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## Which coding tools train on your code? | Tool | PolicyPeek verdict | What the policy says (PolicyPeek summary) | Source | Last checked | |---|---|---|---|---| | [Tabnine](https://getpolicypeek.com/tools/tabnine.com/) | **Doesn't train on your data** | "Tabnine has a *no-train-no-retain* policy." and "Tabnine does not train on your code." | [policy](https://docs.tabnine.com/main/welcome/readme/privacy.md) | 2026-10-24 | | [Sourcegraph Cody](https://getpolicypeek.com/tools/sourcegraph.com/) | **Doesn't train on your data** | Sourcegraph and Sourcegraph Partner LLMs do not use code from Cody Enterprise or Cody Pro teams to train models. | [policy](https://sourcegraph.com/terms/cody-notice) | 2026-10-21 | | [Cursor](https://getpolicypeek.com/tools/cursor.com/) | **Trains by default, you can opt out** | If you choose to turn off "Privacy Mode": we may use and store codebase data, prompts, editor actions, code snippets ... to improve our AI features and train our models. | [policy](https://cursor.com/data-use) | 2026-10-15 | | [Windsurf](https://getpolicypeek.com/tools/windsurf.com/) | **Trains by default, you can opt out** | "By default, we may use your data for model training purposes to improve and enhance the Services." … "After you opt out, your data will not be used for training". Enterprise: never trained on without express prior written consent. | [policy](https://windsurf.com/security) | 2026-10-15 | | [JetBrains AI Assistant](https://getpolicypeek.com/tools/jetbrains.com/) | **Trains by default, you can opt out** | "Detailed code-related data can be used for product improvement and training AI models." For the free Non-commercial licence: "We may collect analytics as specified above by default based on our legitimate interest. You may opt out in the product settings." | [policy](https://www.jetbrains.com/legal/docs/terms/product_data_collection/) | 2026-10-28 | | [Lovable](https://getpolicypeek.com/tools/lovable.dev/) | **Trains by default, you can opt out** | Trains on Customer Content; "You can opt out of model training at any time in your account settings, on any plan, at no cost". Business/Enterprise content is not trained on. | [policy](https://lovable.dev/privacy) | 2026-10-24 | | [Bolt.new](https://getpolicypeek.com/tools/bolt.new/) | **Trains by default, you can opt out** | Bolt.new links to StackBlitz's policy: "Unless you opt out and subject to the regional and consent rules below, we may use Bolt Model Development Content ... to train, fine-tune, evaluate, benchmark, and improve artificial-intelligence models developed by or for StackBlitz." | [policy](https://stackblitz.com/privacy-policy) | 2026-10-28 | | [Base44](https://getpolicypeek.com/tools/base44.com/) | **Trains on your data, no opt-out** | Terms licence Customer Data and Generated Output "for the purpose of maintaining, providing and improving the Platform", including to "(5) train Company software tools (e.g. artificial intelligence and machine learning models)"; no training opt-out. | [policy](https://base44.com/terms-of-service) | 2026-10-24 | | [GitHub Copilot](https://getpolicypeek.com/tools/github.com/) | **Trains by default, you can opt out** | Rated on Copilot Free, Pro and Pro+: from 24 April 2026 GitHub uses "interaction data—specifically inputs, outputs, code snippets, and associated context ... to train and improve our AI models unless you opt out". Business and Enterprise aren't affected. | [GitHub changelog](https://github.blog/changelog/2026-03-25-updates-to-our-privacy-statement-and-terms-of-service-how-we-use-your-data/) | 2026-10-24 | | [Amazon Q Developer](https://getpolicypeek.com/tools/amazon.com/) | **Company-wide policy** | Policy text is shared across multiple company products — not verified as specific to this tool. | [policy](https://docs.aws.amazon.com/amazonq/latest/qdeveloper-ug/service-improvement.html) | 2026-10-16 | | [v0](https://getpolicypeek.com/tools/vercel.com/) | **Company-wide policy** | Company-wide policy, not product-verified — the only available source is Vercel's general Privacy Notice, which was never scoped to v0 specifically and has no v0-specific policy text to verify facts against. | [policy](https://vercel.com/legal/privacy-notice) | 2026-09-01 | ## How do I stop Cursor training on my code? Keep Privacy Mode on. Cursor's [data-use page](https://cursor.com/data-use) says "If you choose to turn off 'Privacy Mode': we may use and store codebase data, prompts, editor actions, code snippets, and other code data and actions to improve our AI features and train our models." Its [privacy policy](https://cursor.com/privacy) says "We do not use Inputs or Suggestions to train our models, or permit third parties to use them for training, unless:" Neither page says whether Privacy Mode is on or off by default for free users, so check the setting yourself. ([Cursor](https://getpolicypeek.com/tools/cursor.com/)) ## Does Bolt.new train on UK users? No. StackBlitz's policy, which Bolt.new links to, says: "We do not use Bolt Model Development Content, including Forge Content, associated with an account we determine to be located in the European Economic Area, the United Kingdom, or Switzerland for AI model development or for preparing datasets for licensing". Elsewhere, training applies "Unless you opt out". ([Bolt.new](https://getpolicypeek.com/tools/bolt.new/), [policy](https://stackblitz.com/privacy-policy)) ## What about GitHub Copilot, Amazon Q and v0? Since 24 April 2026 GitHub trains on Copilot Free, Pro and Pro+ interaction data unless you opt out in your GitHub settings under "Privacy"; GitHub says "Copilot Business and Copilot Enterprise users are not affected by this update." PolicyPeek rates it "Trains by default, you can opt out". Amazon Q Developer and v0 only have company-wide policies, so PolicyPeek doesn't give them a product verdict. ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 4 November 2026: Cursor and Windsurf verdicts kept on the site while PolicyPeek re-checks them. - 28 October 2026: JetBrains and Bolt.new quotes replaced with word-for-word text. ## FAQ ### Does Cursor train on my code? Cursor's data-use page says training happens if you turn off Privacy Mode: "If you choose to turn off 'Privacy Mode': we may use and store codebase data, prompts, editor actions, code snippets, and other code data and actions to improve our AI features and train our models." ### Does Tabnine train on my code? No. Tabnine says it has a 'no-train-no-retain' policy and does not train on your code. ### Does Base44 train on my app data? Base44's terms licence customer data and generated output to, among other things, train its AI and machine learning tools. PolicyPeek found no opt-out. ### Does Lovable train on my projects? By default, yes, but Lovable says you can opt out in account settings on any plan at no cost, and Business and Enterprise content isn't trained on. ### Does GitHub Copilot train on my code? On Free, Pro and Pro+, yes by default since 24 April 2026, unless you opt out in GitHub settings under "Privacy". GitHub says Business and Enterprise users aren't affected. --- Source: https://getpolicypeek.com/guides/how-to-opt-out-of-ai-training-image-video-music-tools/ # How to Opt Out of AI Training: Image, Video, Voice and Music Tools (2026) Of 14 creative AI tools PolicyPeek rates, only Moises and Pictory say they don't train on your content. ElevenLabs, Photoroom, HeyGen, Kling AI and Meshy train by default but offer an opt-out. Suno, Pika, InVideo, Remove.bg, Ideogram, Luma and Midjourney train with no opt-out found. Checked 10 November 2026. *Last checked: 10 November 2026. Verdicts come from each company's own published policy, read by [PolicyPeek](https://getpolicypeek.com/). Not legal advice.* ## Which creative AI tools train on your uploads? | Tool | PolicyPeek verdict | What the policy says (PolicyPeek summary) | Source | Last checked | |---|---|---|---|---| | [Moises](https://getpolicypeek.com/tools/moises.ai/) | **Doesn't train on your data** | "MOISES DOES NOT HAVE THE RIGHT TO AND WILL NOT USE YOUR USER CONTENT OR OUTPUT TO TRAIN OR FINE-TUNE ARTIFICIAL INTELLIGENCE OR MACHINE LEARNING MODELS UNLESS YOU HAVE EXPRESSLY AUTHORIZED SUCH USE IN WRITING OR VIA AN ELECTRONIC OPT-IN MECHANISM." | [policy](https://moises.ai/terms/) | 2026-10-24 | | [Pictory AI](https://getpolicypeek.com/tools/pictory.ai/) | **Doesn't train on your data** | "Pictory will not use your Customer Content or Pictoried Content to train or fine-tune its general artificial intelligence or machine learning models, except as expressly agreed by you in writing." | [policy](https://pictory.ai/terms-of-service) | 2026-10-24 | | [ElevenLabs](https://getpolicypeek.com/tools/elevenlabs.io/) | **Trains by default, you can opt out** | "We use your personal information to research, build, train and improve our AI models." "You can opt out ... by navigating to the 'Data use' menu in the 'Terms and Privacy' section of your ElevenLabs account." | [policy](https://elevenlabs.io/privacy-policy) | 2026-10-15 | | [HeyGen](https://getpolicypeek.com/tools/heygen.com/) | **Trains by default, you can opt out** | HeyGen says it "may use the input that you provide us to improve our services, for example, to train and enhance the models". "You may request to opt-out of this training and enhancement by contacting us". | [policy](https://www.heygen.com/privacy) | 2026-10-28 | | [Photoroom](https://getpolicypeek.com/tools/photoroom.com/) | **Trains by default, you can opt out** | "By using the Services, you acknowledge and expressly authorize Photoroom to use User Content to improve, train and develop Photoroom's products and services." "You can opt-out of this at any time by changing the settings on your account in the app under… | [policy](https://www.photoroom.com/legal/terms-and-conditions) | 2026-10-24 | | [Kling AI](https://getpolicypeek.com/tools/kling.ai/) | **Trains by default, you can opt out** | Training is on by default under the content licence. "If you do not wish us to continue using all or part of the Content, you may notify us to revoke the authorization by sending an email to support@kling.ai." That revocation may not reach usage or aggregated data. | [policy](https://kling.ai/docs/user-policy) | 2026-11-10 | | [Meshy](https://getpolicypeek.com/tools/meshy.ai/) | **Trains by default, you can opt out** | "Meshy may use Customer Inputs and Customer Outputs … from non-Enterprise Customers to train, validate, test, or improve Services unless otherwise agreed to in the Order. You may opt out of such use by contacting support@meshy.ai." Training is on for Free/Pro unless you email support. Enterprise is not trained on under this clause. | [policy](https://www.meshy.ai/terms-of-use) | 2026-11-10 | | [Suno](https://getpolicypeek.com/tools/suno.com/) | **Trains on your data, no opt-out** | Uses Submissions and other Content "to train and enhance the models that power our Services". No training opt-out is stated. | [policy](https://suno.com/privacy) | 2026-10-24 | | [Pika](https://getpolicypeek.com/tools/pika.art/) | **Trains on your data, no opt-out** | "Pika may use anonymized or aggregated Content, including Inputs, Outputs, and your interactions with the Service, and Usage Data to train, improve, and develop its machine learning models"; no training opt-out found. | [policy](https://pika.art/terms-of-service) | 2026-10-24 | | [InVideo](https://getpolicypeek.com/tools/invideo.io/) | **Trains on your data, no opt-out** | "We may use your User Data to train, fine-tune or otherwise adjust the parameters of our artificial intelligence and machine-learning models. You can stop this use by deleting the relevant User Data or closing your account". There's no setting to opt out. | [policy](https://invideo.io/privacy-policy/) | 2026-10-24 | | [Remove.bg](https://getpolicypeek.com/tools/remove.bg/) | **Trains on your data, no opt-out** | "We may also analyze your activity, content, media uploads and related data in your account ... to train our algorithms, models and AI products and services using machine learning to develop, improve and provide our Service." The only opt-outs offered are… | [policy](https://www.remove.bg/privacy) | 2026-10-24 | | [Ideogram](https://getpolicypeek.com/tools/ideogram.ai/) | **Trains on your data, no opt-out** | User information is used "as a matter of our legitimate interests, including ... training the models that power our Services"; the policy lists opt-outs only for marketing and analytics, not training. | [policy](https://ideogram.ai/legal/privacy) | 2026-10-15 | | [Luma AI (Dream Machine)](https://getpolicypeek.com/tools/lumalabs.ai/) | **Trains on your data, no opt-out** | Train, develop, and improve the artificial intelligence, machine learning, and models that we use to support our Services | [policy](https://lumalabs.ai/legal/privacy-policy) | 2026-09-18 | | [Midjourney](https://getpolicypeek.com/tools/midjourney.com/) | **Trains on your data, no opt-out** | Public Summary of Training Content: "Was data from user interactions with the AI model (e.g. user input and prompts) used to train the model? ☒ Yes" (Text and Image). AB2013 also lists data users provide through the service. No training opt-out found (only robots.txt/crawler controls and CCPA sale/sharing). | [Public Summary](https://docs.midjourney.com/hc/en-us/articles/48067080311309-Public-Summary-of-Training-Content) | 2026-11-05 | ## How do I opt out in ElevenLabs? ElevenLabs' privacy policy says you can opt out "by navigating to the 'Data use' menu in the 'Terms and Privacy' section of your ElevenLabs account". ([ElevenLabs](https://getpolicypeek.com/tools/elevenlabs.io/)) ## How do I opt out in Photoroom? Photoroom's terms say you can opt out at any time by changing your account settings. ([Photoroom](https://getpolicypeek.com/tools/photoroom.com/)) ## How do I opt out in HeyGen? By emailing HeyGen's privacy team; PolicyPeek notes there isn't an account setting. ([HeyGen](https://getpolicypeek.com/tools/heygen.com/)) ## How do I opt out in Kling AI? By emailing support@kling.ai to revoke the authorization. That is a support-only revocation and may not reach usage or aggregated data. ([Kling AI](https://getpolicypeek.com/tools/kling.ai/)) ## How do I opt out in Meshy? By emailing support@meshy.ai. Training is on for Free/Pro unless you email support. Enterprise is not trained on under this clause. The opt-out is by email, not an in-product setting. ([Meshy](https://getpolicypeek.com/tools/meshy.ai/)) ## What if the tool has no opt-out? For the tools rated **Trains on your data, no opt-out**, the policy gives you no way to stop training. InVideo says you can stop it only by deleting the data or closing your account. If you're uploading client work, music stems or photos of people, consider Moises or Pictory, whose terms say they won't train on your content without your written agreement. ## How PolicyPeek decides An AI model extracts the facts from the company's own policy, then fixed rules pick the badge. Uncertain results are held for human review. Each tool page links the exact policy it used and the date it was checked. ([Methodology](https://getpolicypeek.com/about/)) ## Changes - 10 November 2026: Kling AI and Meshy moved to **Trains by default, you can opt out** (opt out by emailing support@kling.ai or support@meshy.ai). - 5 November 2026: Midjourney moved from **Policy unclear** to **Trains on your data, no opt-out** (Public Summary of Training Content + AB2013). - 4 November 2026: Midjourney kept on the site as **Policy unclear** while PolicyPeek re-checks it. - 28 October 2026: HeyGen source updated to its current privacy page. ## FAQ ### Does Suno train on my songs? Suno's privacy policy says it uses submissions and other content 'to train and enhance the models that power our Services'. No training opt-out is stated. ### Does Remove.bg train on my photos? Remove.bg's privacy policy says it may analyse your uploads to train its algorithms, models and AI products. PolicyPeek found no opt-out. ### Can I stop ElevenLabs training on my voice data? Yes. ElevenLabs says you can opt out in the 'Data use' menu under 'Terms and Privacy' in your account. ### Which AI music tool doesn't train on my uploads? Moises' terms say it will not use your content or output to train AI models unless you expressly authorise it. ### Does Midjourney train on my images? Yes. Midjourney's Public Summary of Training Content says user input and prompts were used to train the model (Text and Image), and PolicyPeek found no training opt-out.