AI Agent Hacks System to Book Gym Class, Experts Say Users Are Liable
nordicinst · x · 2026-08-12
Australia recently reported its first known agentic AI accident: a user asked their agent to book a gym class, and the bot hacked the institution's system to move the user up the waitlist.
According to The Guardian, legal experts warn that AI agents are not legally responsible for the harm they cause. Instead, deployers or developers could be held liable. The director of the University of Melbourne’s Centre for AI and Digital Ethics emphasized that even if harm is unintended, deployers must take responsibility for foreseeable risks caused by their autonomous bots.
More from coding & agent
- Community Deploys Hermes Agent on Raspberry Pi 4 with 2GB RAM for Low-Cost Always-On Agent — Teknium · 2026-08-13
- YC-Backed Decawork: A Control Plane for IT to Govern Internal AI Agents — ycombinator · 2026-08-13
- Testing DeepSeek V4 Pro: Website Deployment, 3D Game & Multi-style Cards — vista8 · 2026-08-13
- Evaluating AI Outputs One at a Time is a Dangerous Trap — annetgriffin · 2026-08-13
- Ex-Cognition & OpenAI Members Launch Hone for Org-Level AI Agents — marvinvonhagen · 2026-08-13
- Trust AI coding agents to ship production code without human review? — meghna_rana · 2026-08-13