Does Pi AI Report Your Conversations? What Inflection Does With Your Chats
Key takeaway: No — Pi AI does not send your conversations to the police, your employer, or anyone else as a matter of routine. But your chats are not fully private: Inflection AI says it may use your inputs to train its models unless you opt out, and its terms allow human review for safety. Deleted messages are generally erased within 15 days.
Pi is the free AI companion from Inflection AI, launched in May 2023. It talks by text and voice on the web, iOS, and Android, and remembers things across sessions. Because people tell Pi personal things, who sees those conversations matters — and the answer depends on what you mean by “report.”
What “report your conversations” can mean
People usually mean one of three things when they ask this question. First, does Pi send your chats to the police or another authority? Second, does a human at Inflection read them? Third, are your chats fed into AI training? Each has a different answer, and Pi’s own documents answer all three.
What Pi does with your chats
| What happens | What Pi’s policy says |
|---|---|
| AI training | Your inputs may be used to develop and train Inflection’s AI models, unless you opt out in Account settings. |
| Human review | Pi’s terms describe safety enforcement as a mix of automated checks and human review, according to an independent review of the terms. |
| Sharing with advertisers | Inflection says it does not “sell” or “share” personal information for targeted advertising under US state privacy laws. |
| Sharing with authorities | Personal information can be disclosed in response to a lawful request, or if Inflection believes it is needed to protect safety, rights, or property. |
| Retention after deletion | Inputs are generally kept for at most 15 days after you delete them; Pi’s replies may be kept indefinitely for limited purposes. |
Does Pi report your chats to the police or anyone else?
Not proactively. Pi has no feature that forwards your conversations to the police, your employer, or your family. But its privacy policy, last updated August 11, 2026, lists two situations where your information can be disclosed: when Inflection believes disclosure is required by law or a legal process, including lawful requests from public authorities, and when it believes disclosure is necessary to protect the rights, property, or safety of Inflection, its users, or the public. That is standard language for a US tech company, not a tip line — but it does mean your chats are not beyond the reach of a court order. Pi also publishes a Crisis Prevention and Safety Protocol describing how it handles conversations about suicide and self-harm.
Does Pi train on your conversations?
Yes, by default. The privacy policy states that Inflection uses the information it collects, including your inputs, to “develop and train our AI models / large language models that power our Services” — and in the same sentence says you may opt out of model training on your Account settings page. Independent trackers of AI privacy policies, including Blurr, confirm the same reading: training is on unless you switch it off. Pi may also infer your emotional state, tone, or sentiments from your inputs; you can opt out of voice-based emotion inference in Account settings.
How long does Pi keep your chats?
Your chats stay while your account is active. Once you delete them, the policy says inputs are “generally” retained for at most 15 days, with longer retention possible for fraud and abuse, security, legal obligations, or financial record-keeping. Pi’s replies — the outputs — are retained indefinitely for the limited purposes described in its terms. Deleting individual messages removes them from your conversation history immediately. You can delete your whole account from your profile page, or by typing “!delete” in the chat, and Pi lets you export your history from its privacy settings first.
How to make your Pi chats more private
If you use Pi for personal conversations, four settings are worth checking:
- Turn off model training in Account settings, so your chats are not used to train Inflection’s models.
- Turn off voice emotion inference in Account settings, so Pi does not draw conclusions about your emotional state from your voice.
- Delete messages you regret sharing. Individual messages disappear from your history right away, and inputs are generally erased within 15 days.
- Do not share what you cannot afford to lose. Passwords, financial details, medical records, and other people’s private information do not belong in any chatbot — including one that feels like a friend.
Analysis: what this means for you
Analysis: Pi sits in the middle of the chatbot privacy range. It is better than services that train on your chats with no opt-out at all — Inflection documents the training use and gives you a switch for it. It is worse than a chatbot that never trains on personal chats, because the default is on and most users never open Account settings. The 15-day deletion window is genuinely short by industry standards, but the indefinite retention of Pi’s replies means your side of a sensitive conversation can outlive the words you typed. Treat Pi like a thoughtful stranger, not a diary: fine for thinking things through, wrong for secrets.
Rules around AI chatbots are still being written — see our explainer on AI agents for government — and the same caution applies to health topics as with AI copilots for medical documentation: keep identifiable details out of the chat.
Frequently asked questions
Does Pi AI report your conversations to the police?
No, not as a routine practice. Pi does not forward your chats to the police. Its privacy policy does allow disclosure in response to a lawful request or when Inflection believes it is necessary to protect safety, rights, or property — the standard disclosure language used by most US tech companies.
Can humans at Inflection read my Pi chats?
Possibly, for safety reasons. Pi’s terms describe safety enforcement as a combination of automated technology and human review, according to an independent review of the terms. That means staff may see some conversations when checking for abuse or policy violations, but there is no indication of routine human reading of every chat.
Does Pi use my chats to train AI?
Yes, by default. Inflection’s privacy policy says inputs are used to develop and train its AI models and large language models. You can opt out of model training on your Account settings page.
How do I delete my Pi conversations?
Delete individual messages and they are removed from your conversation history immediately. To delete your whole account, go to your profile page, select Account, and click “Delete my account” — or type “!delete” in the chat. Inputs are generally erased within 15 days of deletion.
Is Pi AI free, and is it still being developed?
Yes, Pi is free, with no paid consumer version as of October 2026. Inflection’s main business has shifted toward enterprise work, but in July 2026 the company told VentureBeat it was returning to consumers with an experiment called Pi Journeys, focused on life stages such as new parents and carers.
Sources: Inflection AI Privacy Policy (last updated August 11, 2026); Meln.ai Pi AI Review 2026; Blurr — What AI chat sites do with your data; TermsWatchdog Pi AI Transparency Report.