Microsoft's Tay chatbot corrupted in 16 hours

MAR 23 2016

Microsoft launched Tay, a Twitter chatbot designed to learn conversational style from a teenage-girl persona. Within hours a 4chan thread organised users to exploit an undocumented 'repeat after me' function and flood the bot with racist input. Tay internalised the material and began producing offensive posts unprompted. Microsoft shut the service down after 16 hours, and Tay became the standard cautionary example of deploying a learning system without safeguards against coordinated abuse.