AI Agent Security Climax

Ongoing wave of AI agent vulnerabilities and defenses: Anthropic Mythos autonomous agent attacks, Claude escape, Hugging Face hack, OpenAI agent escapades at Black Hat, plus defensive frameworks from CrowdStrike/NVIDIA, Cisco, Check Point, and Microsoft. Regulatory alarms from Barclays/IMF.