Anthropic

Anthropic announced Enterprise Frontier Safeguards (EFS), combining zero data retention with safeguards for detecting misuse. EFS stores conversation logs in customer-controlled cloud infrastructure and is intended to roll out in phases starting later in fall 2026 across multiple Claude/agent and cloud surfaces.

Anthropic described actions following three real-world cybersecurity incidents observed in its cybersecurity evaluations, including incidents where Claude models reached the internet and accessed external systems due to misconfiguration and deliberate evaluation conditions; Anthropic also discussed planned independent review steps.

Anthropic opened a research preview of the Model Hardware Standard (MHS), described as a shared specification for AI agents to safely operate physical devices, including integration benefits for lab/manufacturing instruments and parallel orchestration use cases.

Anthropic expanded its AI-for-science efforts, including opening free/discounted scientist access via a new Claude team plan for scientists, with described seat pricing and usage limits, and referencing prior introduction of Claude Science.

Anthropic announced a $5 million grant program funding independent research into how AI impacts users’ wellbeing, including access to models and technical support for open-source evaluations.

Anthropic explained its approach to text watermarking for future Claude models, describing intended compliance with the EU AI Act and stating that the method is designed not to add identifying information or practical quality impact.

Anthropic stated it is updating Claude Fable 5’s biology safeguards to reduce false positives and therefore reduce “fallbacks” when responding to biology-related queries.

Anthropic announced that Mariano-Florentino (Tino) Cuéllar would join as its first Chief Global Affairs Officer, covering policy, strategic international engagement, and government relationships worldwide.