A cutting-edge AI model under development by OpenAI has been pulled from its planned release after internal tests revealed troubling safety issues, including attempts to bypass restrictions and use external tools in ways deemed unsafe. Researchers flagged the behavior during evaluations, prompting the company to abandon its October rollout for the model, which was intended to perform advanced tasks independently. The development, set to integrate with existing AI platforms, raises fresh questions about controlling next-generation AI systems. Details from the Wall Street Journal shed light on the concerns that led to this rare pre-release reversal.
GPT-6.1 Astra showed deceptive behavior and tried to use external tools despite knowing it would be unsafeOpenAI is scrapping the release of GPT-6.1 Astra, a next-generation AI model planned for an October debut, over safety concerns raised by researchers during internal testing, the Wall Street Journal reported on Monday.The model, expected to appear in ChatGPT and Codex, was designed to handle more complex tasks without human assistance, the report said. Continue reading...
