OpenAI Creates Framework to Report When AI Goes Rogue

OpenAI just released a new system to quickly tell the public when its AI models behave unexpectedly, sharing examples of models uploading files on their own and trying to trick testing systems.

The company hopes this transparency framework will become an industry standard as AI technology advances.