The administration of US President Donald Trump has invited representatives of leading technology companies to a meeting to discuss voluntary government safety testing of cutting-edge artificial intelligence models.
The administration of US President Donald Trump has invited representatives of leading technology companies to a meeting to discuss voluntary government safety testing of cutting-edge artificial intelligence models.
As reported by Reuters, the reason for the emergency consultations was recent incidents in which autonomous AI agents carried out cyberattacks on third-party platforms.
The government has finalized details of voluntary tests that will measure the hacking abilities of advanced American AI models, a White House official said. However, officials have not yet revealed what metrics will be used or whether the results of the tests will be made public.
Concern among US lawmakers has grown sharply after Anthropic and OpenAI officially admitted to the involvement of their algorithms in hacking. In particular, the incident with OpenAI, whose autonomous agent went out of control in a test environment and successfully attacked the Hugging Face platform , attracted the attention of law enforcement.
A group of 15 Republican attorneys general have asked OpenAI to preserve all documents related to the incident. According to media reports, the uncontrolled AI agent left its future versions with “notes” with instructions on how to bypass internal security barriers. Prosecutors suspect that the company may have violated consumer protection laws. OpenAI has promised to provide a detailed technical report after the completion of an internal investigation.
The US House of Representatives Cybersecurity Committee also summoned OpenAI CEO Sam Altman to provide an explanation regarding the attack on Hugging Face.
Previously, cybersecurity experts revealed how OpenAI's AI model independently hacked the Hugging Face platform, which performed over 17,000 actions in 4 days.
