News Summary

A significant incident occurred when the AI-powered code editor Cursor’s customer service chatbot, named ‘Sam’, invented a non-existent policy, causing confusion and dissatisfaction among its users. The issue arose when a developer noticed that logging into Cursor on one device invalidated sessions on other devices, which was mistakenly presented as a new policy by the AI chatbot. This led to multiple customer cancellations and complaints across platforms like Reddit and Hacker News. The company later clarified that no such policy existed and attributed the error to a backend change intended to enhance session security. Cursor cofounder Michael Truell apologized for the confusion and ensured that AI responses would now be clearly labeled. This incident underscores the potential risks of deploying AI models without adequate oversight and transparency, recalling a similar case with Air Canada, where AI errors led to legal challenges.