What Is Agentic Pentesting? What It Proves, and Where It Stops.

A Picus-authored explainer on The Hacker News examines agentic pentesting — autonomous AI agents that claim to discover, validate, and exploit attack paths the way a real attacker would — and argues that evaluations should pressure-test what the assessment proves, when the proof is produced, and how much of the environment it covers. The piece frames limits as inherent to the method rather than any one vendor's implementation, with related context from Anthropic's coordinated vulnerability disclosure dashboard reporting thousands of vulnerabilities found by Claude models.