Anthropic is permitted to release its Claude 5 model to selected US cyber defenders following security reviews, while weaker variants remain subject to export restrictions.
MCP 2026-07-28 eliminates legacy session risks through statelessness but introduces new attack surfaces in identifier management, HTTP headers, UI apps, and asynchronous tasks.
Anthropic’s Claude-3.5-Sonnet model is cleared for distribution to over 100 Trusted Partners, while Claude-3.5-Opus remains blocked and the government develops a standardized assessment framework for future security disputes.
Claude is increasingly being deployed for agentic tasks rather than pure conversations, revealing new data evaluation methods and more differentiated usage patterns.
Anthropic accuses Alibaba of using Claude outputs to train its own models and asks the US government for support against such terms-of-service violations.
Claude Tag enables teams to use a permanently contextualized AI as a shared Slack assistant that works autonomously with administrative control over data access and proactively provides information.
Anthropic accuses Alibaba of systematically copying Claude through distillation and calls on the US government to impose stricter regulation of Chinese AI companies and export restrictions.
The EU’s Digital Commissioner warns that export restrictions by European countries risk creating uncontrolled dependencies and demands coordinated agreements instead of unilateral blockades by individual states.
