THE DECODER

For the first time, insurers in London’s Lloyd’s market are offering dedicated policies that cover damages caused by errors from AI chatbots. The product was developed by Armilla, a Y Combinator-backed startup, and is designed to protect companies if they get sued over faulty AI performance—such as when customers are harmed by incorrect answers or so-called "hallucinations" from a chatbot. The coverage includes legal fees and compensation payments. A recent example comes from Air Canada, where a chatbot promised a nonexistent discount that the airline later had to honor. According to Armilla, the new policy would have applied in such a case if the bot’s performance fell significantly below expectations. Karthik Ramakrishnan, Armilla’s CEO, says the goal is to make it easier for businesses to adopt AI. Traditional tech insurance often offers only limited coverage for AI-related risks, but Armilla’s policy specifically insures against performance drops in AI models.