Microsoft Security chief says AI 'cannot be trusted' and AI agents are given too much power

Microsoft says 'frontier models, regardless of their origin, cannot be trusted,' and that AI Agents are 'unpredictable' and need supervision.

Published: 2026-10-26