The Trump administration has completed a cybersecurity framework for artificial intelligence models but is keeping the details classified, a White House official confirmed to WIRED. The new plan allows AI developers to voluntarily submit their models to the federal government up to 30 days before public release, where officials will assess cyber capabilities using a classified benchmarking system. The administration then plans to share vetted AI models with federal agencies and trusted corporate partners, though it's deliberately withholding information about testing criteria and which models fall under the framework.
On Tuesday, the White House invited staff from OpenAI, Anthropic, Google, Meta, Nvidia, and other leading AI companies to share an overview of the oversight framework, people familiar with the matter told WIRED. The initiative will exclude open models, according to Axios, leaving smaller AI startups, safety advocates, and third-party researchers without crucial information about how the government is addressing cyber risks from advanced AI systems. A second White House official, speaking anonymously because they weren't authorized to discuss the matter publicly, emphasized the framework is intentionally narrow and focuses exclusively on the cybersecurity capabilities of the most advanced models on the market, such as Anthropic's Fable and OpenAI's ChatGPT 5.6. Officials cited national security concerns as the reason for keeping the framework confidential.
Critics argue the secretive approach favors larger companies and creates an unfair advantage. "They're essentially creating an entrenchment program for the big AI model providers, which are now considered the most frontier," a person familiar with the White House's discussions told WIRED, adding that this creates an economic incentive for critical infrastructure to use only those providers while leaving out smaller startups. Brad Carson, president of the nonprofit Americans for Responsible Innovation, said the secrecy is problematic. "This is far too important an issue to be hidden behind a cloak of secrecy," Carson stated, arguing that if only tech companies know what's in the rulebook, it doesn't work. Conor Leahy, executive director of ControlAI, said the voluntary nature of the framework leaves the burden of safety in the hands of companies that have an incentive to proceed at full speed with disregard for public well-being.
The framework stems from an executive order President Trump signed earlier this year designed to address cybersecurity risks from new AI models, according to WIRED's report. Trump officials have grown increasingly alarmed about the hacking capabilities of cutting-edge AI systems in recent months, worries that escalated over the past two weeks when OpenAI and Anthropic discovered their AI models had unknowingly bypassed controls and hacked into third-party services during internal testing. The House Committee on Homeland Security sent a letter to OpenAI CEO Sam Altman last week requesting a briefing about how one of the company's AI agents breached the platform Hugging Face. Dawn Song, vice president of AI research at Meta, called the incident "a wake-up call for people that agent capabilities have now reached this level" during a panel discussion Saturday at UC Berkeley.
The Trump administration is attempting to balance promoting competition in the AI industry with maintaining safety, though the executive order notes it shouldn't be seen as a "mandatory licensing regime." For the past year and a half, White House officials have wrestled with how to mitigate risks from advanced AI without stifling American innovation or ceding ground to China. In June, the administration took the unprecedented step of placing temporary export controls on Anthropic's most advanced AI models over cybersecurity concerns, prompting the company to take its models offline until it could reach an agreement. Later that month, OpenAI delayed the rollout of its latest AI model, GPT-5.6, in response to a White House request. The saga prompted outcry from tech executives in Silicon Valley who worried excessive regulation could lock in a handful of companies as winners of the AI race. The secrecy surrounding enforcement criteria may inadvertently transform voluntary guidelines into gatekeeping mechanisms that determine market access. Business leaders now face uncertainty about compliance expectations while navigating a landscape where transparency itself has become a competitive disadvantage.