OpenAI shelves new AI model after internal safety tests, WSJ reports - Reuters

OpenAI has decided to halt the rollout of a newly developed artificial‑intelligence model following the results of its internal safety assessments. The decision was reported after the Wall Street Journal disclosed that the company’s own testing indicated concerns that warranted pausing the model’s deployment.

The model in question was engineered to manage more complex tasks autonomously, reducing the need for human intervention in its operations. According to internal documentation, the system was intended to expand the range of capabilities offered by OpenAI’s existing suite of AI tools.

During the safety evaluation, OpenAI’s researchers identified potential risks that could arise from the model’s advanced functionality. While specific details of the identified issues have not been disclosed, the company concluded that further refinement was necessary before any public release.

OpenAI’s move reflects its ongoing commitment to ensuring that its AI technologies meet rigorous safety standards. The company has indicated that it will continue to work on addressing the concerns raised by the internal tests and may revisit the model’s launch once the identified issues are resolved.

The shelving of the model underscores the broader industry focus on responsible AI development, as firms balance rapid innovation with the need to mitigate possible unintended consequences. OpenAI has not provided a timeline for when, or if, the model might be reintroduced after additional safety work.