'K2 Think' AI Model Jailbroken Mere Hours After Release

Researchers discovered that measures designed to make AI more transparent to users and regulators can also make it easier for bad actors to abuse.