Google 将 Gemini 临时对话保留期延长至两个月

作者:Jagmeet Singh原标题:Google to keep Gemini chats for two months, up from 72 hours, even with history off
62AI 研判分

Google 宣布将 Gemini 应用内关闭历史记录后的临时对话保留期从 72 小时延长至约两个月,并将在未来数周全球推送。官方称此举旨在提升 AI 系统的安全与安保能力,但该期限已明显长于 OpenAI ChatGPT 临时对话的 30 天与 Anthropic Claude 隐身对话的 30 天默认值。Google 同时新增独立开关,允许用户单独决定对话是否可用于模型训练。

报道正文

Google is extending how long it keeps temporary Gemini chats and conversations made with chat history turned off to around two months, saying the change aims to improve safety and security as AI systems become more capable.

The change, which will roll out globally in the Gemini app over the coming weeks, marks a significant increase from the previous 72-hour retention period mentioned in Google’s Gemini Apps Privacy Hub, according to an archived version from November 12.

OpenAI, by comparison, says it may keep temporary chats in ChatGPT for up to 30 days, whereas Anthropic retains incognito chats in Claude for 30 days by default. However, companies may keep some data longer for legal or safety reasons, or to investigate policy violations.

Google did not answer questions about what prompted the change and whether it would affect how conversations are accessed or reviewed, instead pointing to a blog post that says chats are kept longer to respond to users, process feedback, and protect its services and users. Its updated Gemini Privacy Hub also cites potential system failures as a reason for retaining chats and says temporary conversations are not used to train AI models.

The longer retention period comes as Google introduces separate settings for Gemini’s chat history and AI training. Until now, both were tied to a single “Keep Activity” setting. Users will now be able to keep their conversations in their chat history without using them to train Google’s generative AI models.

Google says both new settings will be turned off by default for users who previously disabled Keep Activity. For those who had Keep Activity turned on, their chat history will remain enabled. They can, though, decide separately whether Google can use their conversations to train AI models.