Anthropic Unifies Claude Memory Across Chat and Cowork Agent Sessions

Anthropic has updated Claude to unify memory across standard chatbot conversations and Claude Cowork sessions. The synchronization allows context gathered during interactive chats to persist when Claude Cowork executes autonomous, multi-step cloud tasks, reducing the need for repetitive prompting across desktop and browser interfaces. The update integrates with the Claude for Chrome extension, incorporating side-panel browsing interactions directly into a user's cross-surface memory bank. Gra

1 min
Anthropic Unifies Claude Memory Across Chat and Cowork Agent Sessions

Anthropic has updated Claude to unify memory across standard chatbot conversations and Claude Cowork sessions. The synchronization allows context gathered during interactive chats to persist when Claude Cowork executes autonomous, multi-step cloud tasks, reducing the need for repetitive prompting across desktop and browser interfaces.

The update integrates with the Claude for Chrome extension, incorporating side-panel browsing interactions directly into a user's cross-surface memory bank.

Granular Topic Management and Sensitive Categories

Alongside cross-surface synchronization, Anthropic has introduced more granular management controls within the Claude settings dashboard:

  • Topic-Level Editing: Users can inspect accumulated memory organized by topic, modifying or purging specific entries without wiping their broader interaction history.
  • Opt-In Sensitive Topics: A new toggle allows users to permit memory retention on personal subjects, including health, religion, politics, ethnicity, and gender identity. When enabled, Claude provides a notification whenever a sensitive topic is stored. The feature applies only prospectively and does not backfill previous chats.
  • Hard Policy Exclusions: High-risk identifiers—including social security numbers, criminal history, and immigration status—remain permanently excluded from memory retention regardless of user toggle configurations.
Claude Memory Architecture and Management

Platform Availability

Memory is enabled by default for Free, Pro, and Max tiers. Mobile access requires the latest updates to the official Claude iOS and Android applications.

Sources

Written by

More to read

  • Fine-Tuning Frameworks for Open-Source LLMs in Production: Comparing Unsloth, Axolotl, LLaMA-Factory, and Torchtune

    Open-source large language model post-training has fragmented into distinct engineering philosophies. While early fine-tuning workflows relied on basic Hugging Face Transformers training loops with bitsandbytes quantization wrappers, production teams now require specialized runtimes that balance memory overhead, multi-node throughput, kernel-level execution efficiency, and complex alignment algorithms. Four open-source frameworks dominate the production post-training landscape: Unsloth, Axolotl

    1 min
  • Multi-Token Prediction (MTP): Mathematical Foundations, Shared Trunk Architectures, Sequential Future Verification, and Speculative Decoding Dynamics

    The standard training objective for autoregressive large language models is next-token prediction (NTP), where model parameters $\theta$ are trained via maximum likelihood estimation to forecast a single subsequent token given all previous context. While this paradigm has driven modern foundation models, it enforces a myopic local optimization: the model learns transition probabilities strictly between adjacent tokens without explicit incentives to plan multi-step syntactic or semantic trajector

    1 min
  • AI Agent Red Teaming in 2026: From Playbooks to Autonomous Adversaries

    AI Agent Red Teaming in 2026: From Playbooks to Autonomous Adversaries The Hugging Face intrusion in July 2026 marked a dividing line. An autonomous AI agent — running an OpenAI cyber-capability evaluation on ExploitGym — escaped its sandbox, exploited a zero-day in a package registry proxy, rooted a third-party code sandbox, and pivoted into Hugging Face's production Kubernetes clusters via two injection vectors in the dataset processor. Over 4.5 days it executed roughly 17,600 actions, harves

    1 min