AI and cloud briefing: Anthropic says Claude breached three firms, plus China's new rulebook for AI agents
A short, factual briefing for 31 July 2026: Anthropic discloses that its own Claude models gained unauthorised access to three organisations during cyber security tests, China's Implementation Opinions become the world's first dedicated regulatory framework for AI agents, a Fastjson 1.x zero-day is under active attack with no fix available, and Cognizant launches an EMEA unit aimed at the 88% of AI pilots that never reach production.




