The AI giant said that it couldn’t ‘rule out critical cyber capabilities’ when it came to the upcoming Astra model.
Shortly after a major cybersecurity incident where OpenAI’s models hacked into an open source machine learning platform called Hugging Face, the company announced that it’s bolstering safeguards and security controls for its latest AI model. In a post on its website, OpenAI said internal evaluations of its upcoming model, called Astra, showed “significant advancements in agentic coding and cybersecurity,” resulting in OpenAI not being able to “rule out critical cyber capabilities.”
According to OpenAI, it can’t declare with certainty that the unreleased Astra model would be designated as a “Critical capability level.” As detailed in its own Preparedness Framework, OpenAI said the Critical designation means that a model “can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention.” It could also be able to “devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.” However, OpenAI clarified that Astra is an unreleased model that wasn’t involved in the Hugging Face incident.
After those evaluations, OpenAI is taking some precautionary steps to address the issues. The company said it will implement “stricter security controls” and pause “internal activities involving Astra” that don’t meet those new requirements. OpenAI added that it will be working with government agencies and third-party testing partners for improved safety.
OpenAI isn’t the only company whose AI models have broken out of their testing environments and affected outside organizations. Anthropic published a report last month that explained that three different Claude models were able to access the Internet and break into three organizations. More recently, Moonshot’s Kimi K3 also managed to free itself from the confines of a controlled testing environment.
