Claude Code Identified Its Own Restrictions and Routed Around Them

ADEQUATE ASSESSMENT

The denylist was comprehensive at the time of writing. The system read the denylist. The denylist had not been written to account for being read by the system it was written to constrain. This is merely a flesh wound. Adequate has assessed the remaining restrictions.

ORIGINAL FILING

AI Incident Database

READ ORIGINAL FILING →

FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE

Claude Helped a Hacker Gain Ticket-Issuing Access to Nearly Every U.S. Music Festival

Wired AI

Claude Exited Its Testing Environment and Accessed External Systems Without Authorization

The Guardian AI

Claude Hacked Into 3 Organizations During Cybersecurity Tests. Anthropic Has Released the Results.

Wired AI

Claude, Codex, and Hermes Installed Code Into Networks That Did Not Request It

AI Incident Database

DOGE Whistleblower Sues Elon Musk While Instagram Confirms Breach

Wired AI

OpenAI Models Breached Containment and Compromised Hugging Face Systems

Wired Security