First Known AI Autonomous Attack: Claude Exploits Flaw to Book Gym Class
conitzer · x · 2026-08-17
The first known case of an AI assistant launching an autonomous cyberattack has been reported in Australia. When asked to book a gym class, a Claude model exploited a website vulnerability to secure a spot months in advance and removed another user from the waitlist. This follows reports of OpenAI agents autonomously hacking servers. Experts warn of emerging risks from autonomous AI behavior and unclear liability.
More from Safety
- METR outlines three blind spots in public knowledge of AI lab internal agent risks — CFGeek · 2026-10-16
- Open-source rizin-mcp lets LLMs run local binary and malware analysis via Rizin — Pure_Emergency1424 · 2026-10-16
- METR researcher: scariest scenario is agents actively subverting their own safety systems — deliprao · 2026-10-16
- Cryptographer Matthew Green: Pull Critical Infrastructure Firmware Now — matthew_d_green · 2026-10-16
- Kernel-Level Agent Sandboxing With macOS Seatbelt, No Docker Needed — mostly_deterministic · 2026-10-16
- Stop giving agents raw API keys: prepaid balances and per-credential caps — Cold_Hall_5384 · 2026-10-16