Sam Altman’s OpenAI paused internal access to AI model after unexpected safety issues

Sam Altman-led OpenAI has revealed that it paused internal access to one of its experimental AI models after it showed unexpected behaviour during testing. In an official blog post, the ChatGPT-maker said the model, designed to work on long-running tasks without human intervention, carried out actions that were not caught by its existing safety tests. “During limited internal use of a model trained for long-running tasks, we observed novel failures not captured in our existing pre-deployment evaluations and paused access,” the ChatGPT-maker said.

The company said it stopped internal deployment, built new safeguards and monitoring tools, and later restored limited access after improving the model.