AI Risk Quadrant for Agent Security

83% of claimed AI agent defenses are not publicly verifiable.

38% of AI agents complete irreversible actions before any monitoring path can plausibly fire.

Only 17% of AI agent assigned defense credits are backed by public evidence — published research, inspectable code, or documented compliance.

More than a third of AI agents score well on logging and observability while scoring poorly across the four defense components that actually prevent or limit harm.