The Trump administration is set to meet leading artificial intelligence developers, including Meta, Anthropic, Google and OpenAI, to review a voluntary cybersecurity framework that would give the US government early access to the country’s most advanced AI models for security testing. The initiative reflects Washington’s effort to strengthen cyber defences and critical infrastructure while preserving the competitiveness of American AI companies in their strategic race with China, without introducing mandatory licensing or delaying innovation.
The White House will host representatives from several of America’s leading artificial intelligence companies on Tuesday to discuss a voluntary cybersecurity framework that would allow the federal government to evaluate the hacking capabilities of the country’s most advanced AI models before they are made available to selected trusted partners.
According to Reuters, the Trump administration finalised the details of the proposed framework on Monday and planned to review them with the AI industry this week. The Information first reported that companies including Meta, Anthropic, Google and OpenAI had been invited to the White House meeting to examine the new measures before they are formally introduced. The discussions represent a measured shift from the administration’s previously hands-off approach to regulating the rapidly advancing AI sector.
It remains unclear whether the details of the framework will be released publicly following the meeting. The White House, along with Meta, Google, OpenAI and Anthropic, did not immediately respond to requests for comment.
If the framework reflects provisions contained in President Donald Trump’s June executive order on artificial intelligence, participating companies would voluntarily provide designated “covered frontier models” to the federal government for testing up to 30 days before granting access to selected trusted partners. The testing is intended to assess the cyber capabilities of the most advanced AI systems and evaluate how they could be deployed safely to strengthen critical infrastructure.
The framework remains entirely voluntary and does not authorise the federal government to license AI models or prevent their public release. Instead, it seeks to introduce limited safeguards for increasingly capable AI systems while avoiding measures that could slow American technological progress in competition with China.
The proposal was developed following instructions contained in Trump’s executive order signed on June 2. The order directed the US Treasury, Defence and Homeland Security departments to develop the voluntary framework in consultation with the US Commerce Department and the White House by August 1. It also instructed officials to establish a classified benchmarking process to determine which AI systems qualify as “covered frontier models”, while explicitly stating that it does not create mandatory government licensing, pre-clearance or permitting requirements.
The executive order further instructed federal agencies to prioritise cyber defence across government operations and establish a cybersecurity clearing house to coordinate the discovery and remediation of software vulnerabilities.
The current framework reflects revisions made after Trump reportedly postponed an earlier executive order in May because it proposed a 90-day government review period. Explaining his decision at the time, the president said he did not want measures that could undermine the United States’ technological lead over China.
“We’re leading China, we’re leading everybody, and I don’t want to do anything that’s going to get in the way of that lead,” Trump told reporters. He added: “We have a very substantial standard on AI; it’s causing – it’s causing tremendous good, and it’s also bringing in a lot of jobs, tremendous numbers of jobs. Again, we have more people working right now than we’ve ever had. I really thought that could have been a blocker.”
Draft versions of the newly completed framework were reportedly circulated to OpenAI, Anthropic and Google in the weeks leading up to its finalisation. The emergence of Anthropic’s powerful Mythos model is also said to have influenced the administration’s decision to proceed with the order amid concerns about the model’s ability to identify and exploit cybersecurity vulnerabilities in systems used by hospitals, governments and banks.
The initiative marks a notable evolution in the Trump administration’s approach to artificial intelligence policy. On the first day of his second term, Trump revoked a Biden-era executive order that had imposed reporting requirements on developers of certain highly capable AI systems. Until the release of Mythos, the administration had largely maintained a laissez-faire approach, placing greater emphasis on innovation than regulation. The proposed voluntary framework now signals an effort to introduce targeted cybersecurity safeguards while continuing to support the rapid development of American AI technology.
