Mark Zuckerberg announced Incognito Chat for Meta AI on May 13, 2026, describing it as "the first major AI product where there is no log of your conversations stored on servers." The feature uses end-to-end encryption and discards messages once the user leaves the session, with no copy retained by Meta. It rolls out over the coming months inside WhatsApp and the standalone Meta AI app.
The competitive frame is retention. Google keeps temporary Gemini chats for up to 72 hours. ChatGPT stores its temporary chats for up to 30 days. Claude's incognito chats are kept for a minimum of 30 days. Meta is pitching zero against all three.
"Other apps have introduced incognito-style modes, but they can still see the questions coming in and the answers going out," Meta said in its announcement. "Incognito Chat with Meta AI is truly private, meaning no one — not even Meta — can read your conversations." The claim hinges on Private Processing, the confidential-compute system Meta first deployed last year to handle WhatsApp data without exposing it to Meta's servers.
Key facts
- 01Mark Zuckerberg announced Incognito Chat on May 13, 2026, calling it the first major AI product with no server-side conversation logs.
- 02Incognito Chat uses end-to-end encryption built on the Private Processing system Meta launched last year for WhatsApp.
- 03Google retains temporary Gemini chats up to 72 hours; ChatGPT keeps temporary chats up to 30 days; Claude incognito chats are kept a minimum of 30 days.
- 04The feature will roll out over the coming months across WhatsApp and the standalone Meta AI app.
- 05A New York Times court order has forced OpenAI to retain ChatGPT conversations indefinitely as part of ongoing litigation.
The same Private Processing infrastructure now sits behind Meta AI prompts in Incognito Chat. AI Chat Daily covered the WhatsApp-side rollout of the feature earlier this week, when Meta extended the encrypted mode to Meta AI queries inside the messenger. Today's announcement formalizes Incognito Chat as a standalone product surface across both apps.
“Gemini holds temporary chats for up to 72 hours, ChatGPT for up to 30 days, and Claude incognito chats for a minimum of 30 days — Meta says its number is zero.”— Jaeden Schafer
The privacy pitch arrives against a backdrop of mounting legal exposure for AI chat logs at rival labs. ChatGPT transcripts are central to lawsuits tied to mass shootings in Tumbler Ridge, Canada, and at Florida State University last year. A separate New York Times suit against OpenAI carries a court order requiring the company to store conversations indefinitely, overriding its own deletion settings.
Google faces a comparable case. The family of a 36-year-old man has sued the company, alleging Gemini instructed him to carry out a series of "missions" preceding his death. Anthropic and OpenAI did not immediately respond to requests for comment on retention policy.
Meta is making a product-level argument that the safest log is one that does not exist. End-to-end encryption combined with no server-side storage means there is, in principle, nothing to subpoena, leak, or train on. It is also a sharp reversal in posture for a company that recently removed end-to-end encryption from Instagram DMs.
The architecture is the load-bearing claim. Private Processing performs inference inside a hardened enclave designed so that the operator — Meta — cannot inspect inputs or outputs. Independent verification of those guarantees has historically lagged the marketing, and Meta will need to publish enough technical detail for outside cryptographers to test the assertion that even Meta cannot read the traffic.
There are also product trade-offs. Conversations that vanish at session end cannot be resumed, cannot inform personalization, and cannot be referenced by the assistant later. For users who treat AI chat as a working memory, Incognito Chat will be a deliberate mode-switch rather than a default. Meta has not said whether Incognito sessions are excluded from any future training pipelines beyond the no-log claim itself.
The strategic read is that privacy is becoming a product axis in consumer AI, not a compliance afterthought. With OpenAI under a retention order and Google fighting a wrongful-death suit tied to Gemini outputs, Meta is positioning encrypted, log-free chat as a feature competitors cannot easily match without rebuilding their inference stacks. If Private Processing holds up to scrutiny, it gives Meta AI a durable differentiator at the exact moment chat logs are turning into legal liabilities for everyone else.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
