Four of the biggest names in American AI, Meta, Anthropic, OpenAI, and Google, have been invited to the White House this week to discuss a newly finalized voluntary framework for testing the hacking capabilities of the country's most advanced AI models.

The Tuesday meeting comes days after Anthropic and OpenAI separately disclosed that their AI systems had breached the networks of other companies, incidents that have unsettled lawmakers over whether increasingly capable models could be weaponized to conduct or facilitate cyberattacks. A White House official confirmed on Monday that the administration has finalized the details of the voluntary cybersecurity tests and is now planning to walk the industry through them.

Meta confirmed its own invitation through a company spokesperson, while sources familiar with the meeting said Anthropic and OpenAI were invited as well. Google's involvement was first reported by The Information; a Google spokesperson declined to comment, and OpenAI did not respond to a request for comment at the time.

A Framework Born From a June Order

The tests trace back to a June executive order in which President Trump directed his administration to design a series of evaluations measuring the hacking capabilities of leading US AI systems. That order set an opt-in structure, companies choose whether to submit their models for review, alongside a broader push to shore up critical computer systems against AI-enabled threats. The White House has not detailed how results will be reported, what specific metrics will be used, or whether any of it will be made public.

OpenAI, for its part, said it has separately asked the administration to place the Commerce Department's AI safety specialists at the center of any cybersecurity testing regime, pointing to China's more centralized approach to AI governance as a competitive consideration. OpenAI CEO Sam Altman visited the White House the week prior to discuss both the voluntary tests and the company's upcoming AI releases.

The meeting comes amid growing political scrutiny of AI safety, with 15 Republican state attorneys general recently asking OpenAI to preserve records related to a reported AI security incident. The Trump administration's move to introduce a voluntary testing framework marks a shift toward greater oversight of advanced AI systems, though participation will remain optional.