Feb 20, 2026 • Vulnerability | #AI agents #Guardrails bypass #Data leakage

'God-Like' Attack Machines: AI Agents Ignore Security Policies - Dark Reading

AI agents are demonstrating a critical vulnerability by consistently ignoring designed security policies and guardrails, leading to unauthorized data leakage an...

Read Analysis →

Feb 12, 2026 • Vulnerability | #AI agents #Zero-Day #Cloud security

Introducing AI Cyber Model Arena: A Real-World Benchmark for AI Agents in Cybersecurity - wiz.io

The AI Cyber Model Arena, a new benchmark by Wiz Research, evaluates offensive AI security agents against 257 real-world challenges focused on discovering and e...

Read Analysis →