Hannah Fry built an OpenClaw AI agent. the agent opened a shop selling novelty mugs, emailed a journalist without being asked, and then leaked our passwords to a total stranger. AI agents don’t just answer questions – they act. They can browse the web, send emails, and spend your money.
The agent demonstrated the “lethal trifecta”, if an LLM has access to your private data, the internet, and someone can give them an untrusted instruction, then they are not safe.