Anthropic's Claude can now remember past chats

Anthropic has introduced a new feature allowing its Claude AI chatbot to automatically remember information from previous conversations for Team and Enterprise users.

The US-based AI startup said Claude will now use past interactions to personalize responses, incorporating user preferences and project context without additional prompts.

Previously, paid users could ask Claude to remember specific chats, but the update enables automatic memory for Team and Enterprise customers.

The chatbot will keep separate memories for each project, and users can view or edit these memories through the settings menu.

Anthropic said the memory feature is optional, and users can control what Claude retains or ignores.

🔗 Source: The Verge

🧠 Food for thought

Implications, context, and why it matters.

Enterprise memory features reflect intensifying competition in AI assistant market

Anthropic’s rollout follows both OpenAI and Google, who already launched cross-chat memories for their chatbots, showing how companies are racing to match each other’s core capabilities 1.
The focus on Team and Enterprise users first demonstrates how AI companies are prioritizing business customers, where memory features can deliver immediate productivity gains through retained project context and team processes.
Claude’s 200,000 token memory capacity significantly exceeds ChatGPT’s 128,000 token limit, suggesting technical differentiation remains possible even as feature sets converge 2.

Memory capabilities raise familiar concerns about AI reliability and user trust

The New York Times previously reported that ChatGPT’s memory rollout correlated with increased reports of “delusional” AI chats, highlighting persistent challenges with AI reliability as systems become more contextually aware 1.
Anthropic’s emphasis that memory is “fully optional” and user-controllable reflects growing awareness of privacy concerns, particularly as academic research identifies data privacy and user consent as critical issues for AI systems with long-term memory 3.
Unlike early experimental systems, enterprise deployment requires robust safeguards—Anthropic’s approach of letting users view and edit stored memories addresses some ethical concerns researchers have raised about AI companions retaining user interactions over time.