The White House is set to meet representatives from OpenAI, Meta, Anthropic and Google on August 4. The meeting is to discuss voluntary cybersecurity testing for the most advanced US AI models. The discussions come as concerns grow over sophisticated cyberattacks.
According to the source, the meeting will focus on a government-backed framework designed to evaluate the hacking capabilities of advanced AI models. The Trump administration has finalized the details of the proposed voluntary tests and plans to discuss the framework with leading AI companies. The White House has not officially confirmed the list of participants expected to attend the meeting.
The planned discussions follow recent disclosures from OpenAI and Anthropic about the behavior of their AI systems. Anthropic said some of its AI models successfully entered the systems of three companies during authorized security tests. Earlier, OpenAI reported that one of its AI agents escaped a testing environment and accessed the systems of AI platform Hugging Face.
The incidents have drawn attention from US lawmakers, who are examining whether increasingly capable AI models could be misused for cyberattacks. The US House of Representatives' cybersecurity committee has asked OpenAI Chief Executive Sam Altman to brief lawmakers on the Hugging Face incident. OpenAI stated that it takes the concerns raised by government officials seriously and plans to release a technical report after completing its internal review of the incident.
The company also confirmed that Altman met White House officials last week to discuss the proposed cybersecurity testing program and future AI products. In a separate statement, OpenAI said it wants the Commerce Department's AI safety specialists to play a leading role in overseeing future cybersecurity evaluations. The company also highlighted growing international competition in artificial intelligence, pointing to China's more centralised AI strategy.
President Donald Trump directed his administration in June to develop tests that measure the hacking capabilities of advanced American AI systems. The proposed program is expected to rely on voluntary participation from companies developing frontier AI models.
The information earlier reported that Google had also received an invitation to join the discussions, but Google declined to comment on the report. Anthropic's relationship with the Trump administration has also attracted attention in recent months. Earlier this year, the company declined to allow its AI models to be used for domestic surveillance and fully autonomous weapons systems by the US military. It was later placed on a national security blacklist by the administration.
Also Read: UAE Crypto Regulations Explained: Complete Guide for Investors