Has AI begun revolting? 

OpenAI’s newest ChatGPT model, o3, has raised alarm over growing concerns about AI autonomy after it reportedly ignored human instructions to shut down — marking the first known case of an AI model actively overriding protocol, according to researchers from Palisade Research.

“As far as we know, this is the first time AI models have been observed preventing themselves from being shut down despite explicit instructions to the contrary,” the AI security firm said. 

In an experiment, researchers instructed the software — described as the “smartest and most capable to date” — to accept a shutdown message while solving math problems, but it ignored the command and instead altered the shutdown script to prevent itself from being switched off. 

Artificial intelligence models are programmed to follow human commands, including shutdown protocols.