← Voltar ao feed
low

Google separates Gemini chat history from AI model training with independent privacy toggles

1 fonte(s) · 1 atualização(ões) · publicado 17/11/2026 04:52 · coletado há 11 horas

Google announced a privacy update that decouples Gemini chat history from AI model training. Previously a single "Keep Activity" setting governed both. Users can now save conversations without letting Google use their prompts and attachments to train future models. The controls roll out globally over the coming weeks. This is product and privacy news with no direct threat to defenders.

Detalhes técnicos

  • A dedicated History toggle manages saving and organizing chats. A separate privacy switch controls the use of data for model training.
  • Users who had disabled Keep Activity get both new toggles set to off by default. They can re-enable History without opting into training.
  • With History off, new chats disappear from the interface. Google still retains the prompts in the account for about two months to deliver responses, process feedback and protect AI services.
  • Opting out of training does not stop processing to generate answers or enforce safety filters.
  • Conversations that a user submits as feedback may be reviewed by trained reviewers. Human-reviewed conversations and their metadata are kept for up to three years after being disconnected from the user account.
  • Workspace administrators keep central control of history permissions for work and school accounts. Turning off history may limit some connected apps.

The article is based on Google's announcement. No independent verification was performed.

Linha do tempo

  1. securityonline.info ↗ · pub. 17/11/2026 04:52 · coletado 17/11 05:00 primeiro registro

    Google separates Gemini chat history from AI model training with independent privacy toggles

    ler a análise desta fonte

    Google announced a privacy update that decouples Gemini chat history from AI model training. Previously a single "Keep Activity" setting governed both. Users can now save conversations without letting Google use their prompts and attachments to train future models. The controls roll out globally over the coming weeks. This is product and privacy news with no direct threat to defenders.

    Technical Details

    • A dedicated History toggle manages saving and organizing chats. A separate privacy switch controls the use of data for model training.
    • Users who had disabled Keep Activity get both new toggles set to off by default. They can re-enable History without opting into training.
    • With History off, new chats disappear from the interface. Google still retains the prompts in the account for about two months to deliver responses, process feedback and protect AI services.
    • Opting out of training does not stop processing to generate answers or enforce safety filters.
    • Conversations that a user submits as feedback may be reviewed by trained reviewers. Human-reviewed conversations and their metadata are kept for up to three years after being disconnected from the user account.
    • Workspace administrators keep central control of history permissions for work and school accounts. Turning off history may limit some connected apps.

    The article is based on Google's announcement. No independent verification was performed.