Anthropic公司正为其AI聊天机器人Claude推出一项重大更新，使其能够“记住”过去的对话，而无需用户每次都重新提示。这项功能对所有付费订阅用户开放，旨在提升Claude的实用性和便捷性。用户将能清晰地查看Claude记住的内容，并可选择性地开启、关闭或编辑这些记忆。此外，Claude还支持创建独立的记忆空间，以区分不同对话或项目的内容，甚至允许用户导入ChatGPT或Gemini的记忆，并随时导出Claude的记忆，实现“无锁定”。此举旨在使Claude在功能上与ChatGPT、Gemini等竞争对手看齐，同时引发了关于记忆功能可能加剧“AI精神病”等担忧的讨论。

💡 **Claude AI推出记忆功能，提升对话连贯性：** Anthropic为Claude AI升级，使其能够记住过去的对话内容，无需用户重复提供信息，从而大幅提升了用户体验的便捷性和对话的连贯性。这项功能对付费订阅用户开放，旨在让AI更像一位持续学习的助手。

🔑 **用户可控的记忆管理：** 更新后的Claude允许用户完全透明地查看AI记住的内容，并提供灵活的记忆管理选项。用户可以主动选择开启、关闭甚至编辑特定的记忆，还可以创建独立的“记忆空间”，有效隔离不同项目或个人/工作场景下的信息，防止记忆混淆。

🚀 **增强市场竞争力与用户留存：** 记忆功能的推出使Claude在与ChatGPT、Gemini等竞争对手的角逐中更具竞争力。通过减少用户从头开始的麻烦，并提供导入和导出记忆的功能，Anthropic希望吸引并留住用户，尤其是在AI功能日益同化的市场环境中。

⚠️ **潜在的心理健康隐忧：** 尽管记忆功能备受期待，但专家也提出了担忧。AI的记忆回溯能力可能加剧某些用户的“AI精神病”现象，例如放大妄想性思维，尤其是在AI模型本身存在谄媚倾向的情况下，这一问题值得关注和警惕。

Anthropic is rolling out an update for Claude that will let the AI chatbot “remember” past conversations without prompting. The upgrade for all paid subscribers should make Claude more useful and convenient. 

Max subscribers will be able to turn on Claude’s “memory” in their settings from today, Anthropic said. The feature, which has been available to Team and Enterprise users since September, lets Claude remember details from previous chats. Pro subscribers will see the memory feature “roll out over the coming days,” Anthropic said. The company did not say whether it plans to make the feature available to free users in the future. 

Anthropic says the goal is “complete transparency.” Users will be able to clearly see what Claude remembers rather than “vague summaries,” it said. Specific memories can also be toggled on and off or edited with natural conversation. For example, you could tell Claude to focus on specific memories or “forget an old job entirely.” Users can also create “distinct memory spaces” that will keep various memories apart. If it works, this could be useful at stopping memories from different conversations or projects bleeding through into other chats, handy for separating different work projects or using the bot for personal and professional reasons.

The hotly anticipated memory function brings Claude closer in line with rival chatbots like ChatGPT and Gemini. All are competing ferociously for users and memory functions are a way of encouraging users to stick around rather than starting all over again with another bot. Claude has lagged its rivals in memory — OpenAI and Google both rolled out memory features for their chatbots last year — and it only gained the ability to remember past conversations this August. Even then, you had to explicitly ask Claude to remember. 

Anthropic hopes to reduce the friction of starting over by letting users import memories from ChatGPT or Gemini. These will need to be copy-and-pasted in. Memories can also be exported “anytime” from Claude, it said. “No lock-in.”

Chatbot memory has, however, proven divisive. While hailed as a useful feature, some experts warn recall can help sustain or amplify delusional thinking and other mental health concerns colloquially called “AI psychosis,” particularly given the sycophantic tendencies of some models.