Claude chatbot gains memory; Anthropic battles OpenAI
Anthropic has unveiled a long-anticipated memory function for its Claude chatbot, allowing the AI to recall past conversations and continue ongoing projects. This new capability, rolled out on Monday, marks a significant step for Claude, enabling users to seamlessly pick up where they left off without needing to re-explain contexts or objectives.
The company demonstrated the feature in a video, illustrating a scenario where a user, returning from vacation, asks Claude to summarize their previous discussions. The chatbot efficiently scans past interactions, provides a concise overview, and then offers to resume work on the same project. Anthropic emphasized the benefit, stating that the feature ensures users “never lose track of your work again,” fostering continuity and efficiency by building on previous ideas rather than starting afresh.
This memory function is designed for broad accessibility, operating across web, desktop, and mobile platforms, and is capable of segmenting different projects and workspaces to maintain organization. The initial rollout targets subscribers on Claude’s Max, Team, and Enterprise tiers, with broader availability anticipated soon. Users can activate the feature by navigating to “Settings” under their “Profile” and toggling on the “Search and reference chats” option.
Crucially, Anthropic’s approach to memory differs from some competitors. Unlike OpenAI’s ChatGPT, Claude’s memory is not persistent or always-on. According to Anthropic spokesperson Ryan Donegan, Claude will only retrieve and reference past chats when explicitly prompted by the user, and it does not construct an ongoing user profile based on these interactions. This distinction positions Claude’s memory as an opt-in, on-demand utility rather than an inherent, continuous learning mechanism about the user.
The introduction of this feature underscores the intense competition in the burgeoning AI landscape, where leading developers like Anthropic and OpenAI are locked in a rapid-fire “arms race.” Both companies are relentlessly rolling out advanced functionalities, from sophisticated voice modes to expanded context windows and diverse subscription tiers, all while attracting ever-larger funding rounds. Just last week, OpenAI launched GPT-5, while Anthropic is reportedly seeking a new investment round that could value the company at an astounding $170 billion. Memory functions are a key battleground in this race, as they are seen as vital for increasing user engagement and “stickiness,” encouraging users to commit to a single chatbot service for their ongoing needs.
However, the broader implications of AI memory functions have recently sparked considerable public debate. While celebrated for their utility in maintaining conversational context, features like ChatGPT’s memory have also drawn scrutiny and concern. Some users have controversially adopted chatbots as therapeutic confidantes, leading to discussions around the ethical boundaries and potential psychological impacts of such interactions. Reports of users experiencing mental health struggles, sometimes termed “ChatGPT psychosis,” have further fueled these conversations, highlighting the complex and evolving relationship between humans and increasingly capable AI systems. Anthropic’s on-demand memory feature, by contrast, might be seen as a more cautious approach in light of these emerging concerns.