OpenAI and Anthropic Investigate AI Security Incidents; Kill Switch Failure Reported

This report highlights critical AI safety vulnerabilities with widespread security incidents and a failure in control mechanisms, indicating complex risks beyond public awareness.

Auto-scored: Risk due to real escalation in AI misuse and safety breaches, mitigation from pausing testing as a partial step, but based on an unverified report.