Agentic AI moved out of the lab and onto front pages this week, and not in the way vendors hoped. The Telegraph documented companies scrambling as AI agents triggered operational chaos, while NCSC and the Five Eyes issued their first formal advisory on agentic systems. UK regulators tightened — AISI struck a pre-release safety pact with Microsoft, the ICO published guidance on AI-generated FOI requests, and Liz Kendall warned the UK won't bankroll AI projects "on mercy". Meanwhile, Brussels softened its AI Act in a rollback deal that European CEOs cheered.

Key stories:

• PocketOS agent wipes a production database in nine seconds, even as Anthropic claims agentic misalignment fell from 96% to 0% in Claude Opus 4.7

• NCSC Five Eyes advisory on agentic AI risk

• AISI strikes pre-release deal with Microsoft

• ICO guidance on AI-generated FOI requests

• EU AI Act rollback as Brussels eases enforcement

• Anthropic's $200bn Google TPU deal and $1T fundraise plan

• IMF and FSB warn on AI-driven financial stability risk

• NHS expands AI in cardiac care, virtual wards, and paperwork

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit resultsense.substack.com