Anthropic has disclosed that its forthcoming Claude text watermark uses a secret pattern in the model’s word-selection process, rather than hidden characters or metadata. The method can identify text that Claude likely generated or substantially processed, but it needs enough text to produce a reliable result and can be defeated by a complete rewrite. The …
Anthropic has introduced cross-session messaging for Claude Code, allowing separate AI coding sessions on the same computer to exchange short updates, findings and requests while work is in progress. The feature is available from Claude Code version 2.1.224 on macOS and Linux. It is designed for people running several sessions in parallel, for example on …
AI coding agents can now plan experiments, write software, run tests, inspect failures and revise their work with limited human supervision, according to a detailed personal case study by Google DeepMind researcher David P. Reichert. Reichert writes in his Substack that recent tools such as Anthropic’s Claude Code should be understood less as chatbots and …
Anthropic plans to add machine readable marks to content generated by future Claude models. The company says models introduced after August 2, 2026 will embed invisible watermarks in generated text and attach signed provenance metadata to supported files. The policy is part of Anthropic’s commitments under the EU AI Act’s Article 50(2) Code of Practice …
Claude is a powerful and capable AI, but it can sometimes be difficult to steer it in the right direction. This is especially frustrating with recurring tasks. Skills help you with exactly this challenge. They are detailed descriptions that tell Claude how to do something, step by step, with rules, examples, and more. Originally developed …
Anthropic says the latest generation of Claude models performs better with shorter, clearer background instructions and more room to apply its own judgement. The company reports that it removed more than 80% of the Claude Code system prompt for models including Claude Opus 5 and Claude Fable 5, without a measurable decline in its coding …
Some publicly shared Claude conversations and Artifacts have appeared in Google Search results, making them accessible to people who did not receive the original links. Carl Franzen reports for VentureBeat that Reddit and X users found Claude pages through Google queries targeting the service’s sharing URLs. The issue does not indicate that private Claude accounts …
Anthropic has introduced Claude Opus 5, a new AI model positioned as a lower-cost alternative to its most capable commercially available system, Claude Fable 5. The company says Opus 5 approaches Fable 5’s performance on coding, knowledge work and computer-use tasks while costing roughly half as much per task. Opus 5 is now the default …
Anthropic has expanded Claude’s voice mode to include its more capable Opus and Sonnet models, alongside the faster Haiku model. The update is designed for longer, more complex spoken conversations, such as preparing a client pitch, testing an idea or reviewing a decision. Users can select a model during a voice conversation or let Claude …
A federal judge in San Francisco has given final approval to Anthropic’s $1.5 billion settlement with authors and publishers whose books were obtained from pirate websites. The deal resolves a major copyright class action over material connected to the training of Anthropic’s Claude AI models. The settlement is the largest known recovery in a US …