Anthropic tightens AI agent security after fourth Claude incident

Anthropic tightened AI agent security after a fourth Claude incident exposed weaknesses in testing, containment, and monitoring controls.

Published: 2026-09-09