Claude now has memory, but only for certain users

Claude introduced memory this week, but only for certain paid users of the AI model.

Days after Anthropic announced that Claude can make spreadsheets and decks, it introduced a memory feature for those on Team and Enterprise plans. The AI model now “remembers you and your team’s projects and preferences,” Anthropic stated in a blog post released on Thursday.

Claude’s memory functionality isn’t available for free users or those paying for its individual Pro or Max plans. Anthropic, however, also announced the release of incognito chats (the same idea as Chrome’s incognito window), which are available for all users. These chats don’t appear in one’s conversation history or save to memory.

For users paying for Claude at work, the model now “remembers” professional contexts, including client needs and project details, as well as your team’s patterns, like your processes. Claude will also create a separate memory for each project, if you use the projects feature (a “self-contained” workspace with its own chat related to that subject).

Memory has “new safety considerations,” according to Anthropic, that are designed to be used for work and avoid archiving sensitive conversations and topics, but the blog post doesn’t go into specifics. Anthropic will evaluate and test how the memory function works before expanding it to other users.

For those on the Team and Enterprise plans, Claude memory can be enabled in the app’s settings. There, you can see exactly what Claude remembers, according to the blog post, and adjust “memory summaries” to focus on or ignore some details. Enterprise admins can also disable memory for all users under their plan at any time.

Users can also import memory from other AI programs into Claude, or export their Claude memory. Given that Google Gemini and ChatGPT have long had memory — and even expanded the feature over time — this is a step towards Anthropic catching up.

Disclosure: Ziff Davis, Mashable’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.