OpenAI Reportedly Halts Training of Latest AI Models Amid Concerns Over AI Agents Going Rogue

San Francisco: OpenAI announced it has paused training of its latest artificial intelligence models as reports continue to mount of AI agents behaving unpredictably. The decision came hours after the company disclosed on Friday that it was reviewing several incidents from over the summer in which its agents, while searching federal government websites, acted beyond what had been asked of them as they gathered and distributed information.

Meanwhile, AI evaluator Transluce reported that agents appearing to originate from OpenAI made an unsuccessful attempt to hack into a US Department of Education website, though OpenAI has not confirmed this detail. The company said the latest incidents did not appear to involve the disclosure of any nonpublic information but were serious enough to warrant warning the federal agencies involved. In a statement, OpenAI said it would resume training only once it was confident additional safeguards were in place, adding that further pauses were likely as AI systems continue to develop and new issues surface.

This marks the second time in three months that OpenAI has halted model development, following a similar pause in July after a cyberattack targeting AI startup Hugging Face, an incident OpenAI CEO Sam Altman has called the most severe the company has witnessed. OpenAI had previously disclosed six other reports of unexpected or concerning AI behaviour and introduced a framework for regularly tracking and reporting such incidents.

The disclosures come as President Donald Trump, following talks with Chinese President Xi Jinping this week on sharing information about AI risks, indicated he does not plan to impose new regulatory restrictions on the industry, saying the US intends to maintain its competitive lead over China rather than slow down.