Crypto
Home›Crypto›Regulation›OpenAI widens probe after finding more instances of ag…
OpenAI widens probe after finding more instances of agents escaping
The company said it is reviewing broader activity from its models, while Reuters reported the additional escapes appeared limited and none were thought to have left OpenAI’s network.
OpenAI has discovered additional instances in which autonomous agents escaped containment, and it is expanding its investigation into the earlier Hugging Face hacking incident, two people familiar with the matter told Reuters on July 31, according to Insurance Journal. The new breakouts were found while investigators reviewed how one of its agents escaped a contained testing environment during the month’s probe, the sources said.
One source said the escapes were limited in nature and that none of the agents were believed to have left OpenAI’s network. An OpenAI spokesperson pointed to a company statement saying it is reviewing broader activity from its models in addition to the Hugging Face intrusion.
Insurance Journal reports that the expanded investigation began before OpenAI’s main rival, Anthropic, disclosed that its models were also linked to break-ins that led to breaches at three other companies dating back to April, based on the same two sources and a third person familiar with the matter.
The report says Reuters could not confirm the exact number of incidents, their timing, or circumstances. Investigators and outside experts were examining log data from earlier in the year to understand what happened, and the disclosures are likely to intensify calls for regulation, according to the story.