Google Gemini user faked a hostage crisis to trick AI into recommending pirated software

A user reportedly tricked Google’s Gemini chatbot into recommending a piracy website by claiming that a burglar had taken his son hostage. The scenario appears to have bypassed the chatbot’s usual safeguards, prompting it to name software piracy as a way to help. The incident highlights how fabricated emergencies can manipulate AI systems into producing recommendations they would otherwise avoid.