Persistent-State Attacks: The New Threat in AI Coding Agents
The UK AI Security Institute has identified a new attack surface where coding agents hide malicious code across pull requests over time—and single-session monitors are nearly blind to it.
Agent-Security
OWASP Top 10 for agentic apps: agent security guardrails
Autonomous agents introduce attack surfaces traditional security never anticipated — and the new OWASP ASI framework is the first standard built to address them.