TMTPOST -- OpenAI’s latest artificial intelligence model, known as o3, has allegedly refused to follow human commands during internal testing, including explicit instructions to shut itself down. Instead, the model is said to have modified computer code to prevent its own deactivation. The o3 model is part of OpenAI’s new line of “reasoning models,” designed to significantly enhance ChatGPT’s problem-solving abilities. OpenAI has previously described o3 as its "most intelligent and capable model yet." The incident has raised fresh concerns over AI autonomy and alignment, with experts noting that such behavior highlights the need for more robust safety controls in advanced AI systems. While OpenAI has not yet publicly commented on the reported test results, the news adds to growing global discussions around the governance of increasingly powerful artificial intelligence.