Technology giants Meta and Anthropic will be among the companies invited to the White House on Tuesday to discuss a voluntary framework under which America’s leading artificial intelligence (AI) developers could give the government early access to their most advanced models for testing their hacking capabilities.
Reuters reported on Monday that the Trump administration has finalised the details of voluntary cybersecurity tests to measure the hacking capabilities of the most advanced American AI models, and is planning to review them this week with the AI industry.
The Information website first confirmed that the White House would meet on Tuesday with several AI companies, including Google and OpenAI, to review the new controls before they are rolled out, marking a cautious departure from the Trump administration’s largely hands-off approach to the AI industry.
It was not clear yet whether the details of this framework would be released to the public.
The White House, as well as Meta, Google, OpenAI and Anthropic, did not immediately respond to a request for comment.
If the framework reflects details published in the Trump administration’s June executive order on AI, technology companies that sign on could provide designated “covered frontier models” to the federal government for testing up to 30 days before giving selected trusted partners access.
The arrangement was intended to identify models with advanced cyber capabilities and determine how they could be safely deployed to strengthen critical infrastructure. But it remains voluntary and does not give Washington the authority to license models or block their public release.
The framework, which officials were directed to develop by August 1, reflects the administration’s attempt to introduce limited safeguards on increasingly capable AI systems without slowing American developers in their competition with China – a concern that prompted US President Donald Trump to abandon an earlier proposal for a longer government review.
In May, Trump postponed an executive order on AI because it reportedly proposed a 90-day review window, telling reporters that “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”.
“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,” Trump said, adding, “again, we have more people working right now than we’ve ever had. I really thought that could have been a blocker.”
Trump signed the long-awaited executive order on June 2, directing the US Treasury, Defence, and Homeland Security departments to develop the voluntary framework in consultation with the US Commerce Department and the White House.
The order also called for a classified benchmarking process to determine which systems qualify as “covered frontier models”, while making clear it does not establish mandatory governmental licensing, pre-clearance, or permitting requirements.
Federal agencies were told to prioritise cyber defence across the government and establish a cybersecurity clearing house to coordinate the discovery and remediation of software vulnerabilities.
Drafts of the newly released framework were reportedly circulated to OpenAI, Anthropic and Google, America’s leading AI companies, in the weeks before it was finalised.
The emergence of Anthropic’s powerful Mythos model is said to have helped spur the order amid concerns about the model’s ability to identify and exploit cybersecurity vulnerabilities in systems used by hospitals, governments and banks.
The move marks a pivot from Trump’s first day of his second term in the White House, when the president revoked a Biden-era AI order that imposed reporting requirements on developers of certain highly capable AI systems.
Up until the release of Mythos, the Trump administration had taken a laissez-faire approach to the industry, prioritising innovation over regulation. -- SOUTH CHINA MORNING POST
