Anthropic announced that it has updated the memory system in its artificial intelligence assistant, “Claude,” by unifying the memory used in conversations and the “Claude Cowork” feature, allowing the assistant to retain the information it learns in one use and benefit from it in the other.
The update allows users to view the information that Claude keeps in his memory, with the ability to modify or delete it, which facilitates the transition between the research phase and working on projects without the need to re-explain the details to the assistant every time.
Under the new system, “Claude” begins adding topics to his memory during the conversation, instead of waiting for it to end to summarize it, which allows information to be updated and shared between conversations and “Cowork” faster.
Anthropic said that “Cloud” will not, by default, store personal or sensitive information, such as health data, race, religious beliefs, political opinions, and gender identity, unless the user chooses to activate the option to allow sensitive topics to be saved, and will be notified when any of them is saved.
The company confirmed that there is information that “Cloud” will never save, including government ID numbers, social security numbers, criminal history, and immigration status, in addition to information that violates its usage policies.
The new memory feature is available by default to subscribers of the free, “Pro” and “Max” plans via the web, computers and mobile phones, while “iOS” and “android” users must update the application to the latest version.
