Analysis
AI Agent Security Risks: What Happens When Agents Are Manipulated
Publication date not yet evaluated · Added · 3 reports acalvio.com
Page published
Coverage timeline
Why it matters
OWASP's agentic AI framework and deception-based runtime detection give defenders a shared vocabulary and a complementary control for hijacked agents whose malicious intent lives in the sequence of individually-permitted actions rather than any single step.
Acalvio's write-up synthesizes OWASP's agentic AI security guidance (the AI Agent Security Cheat Sheet, Top 10 for LLM Applications, and Top 10 for Agentic Applications 2026), mapping risks like prompt injection (LLM01), excessive agency, and trust chain abuse to runtime detection requirements. It argues that deception — decoy APIs, deceptive credentials, and honeytokens — adds a behavioral detection layer over SIEM/EDR, citing the disclosed GTG-1002 AI-orchestrated espionage campaign as motivation.