The Trump administration is finalizing voluntary standards for frontier AI model releases after months of executive orders, export controls, and forced model shutdowns.
Normally, this should come as hopeful news as AI companies get what they wanted after lobbying Washington since 2025 to preempt an increasing patchwork of government AI regulations.
The federal government moved late last year with an executive order that created an AI litigation task force to challenge state regulations and called on Congress to create a single federal standard.
The government then established a set of requirements, such as disaggregated benchmarking requirements and pre-issuance government review periods. AI companies have seen this mechanism in action next Anthropy was requested To pull two major models offline within days of launch.
What is the current federal framework for AI?
The Trump administration’s AI policy is implemented through three executive orders. The first, signed in December 2025, directs the Commerce Department to evaluate state AI laws and identify those that conflict with federal policy.
As I created Litigation Task Force To sue states whose regulations the administration considers to be too burdensome.
the Second rankingwhich came in June 2026, tasked the National Institute of Standards and Technology (NIST) with building a secret benchmarking process for frontier AI models within 60 days. It also introduced a pre-release evaluation window of up to 30 days from government access before it can share covered border models with trusted partners.
The administration also stated that no new federal regulator for AI will be created, as oversight is done through existing agencies, which are primarily NIST and the Cybersecurity and Infrastructure Security Agency.
Now it is reported that the US government exists Advanced conversations with artificial intelligence companies to finalize voluntary standards for model releases, with an announcement said to be made soon. The standards will set standards for advanced models and clarify who can access them domestically and abroad.
The US government has had a hand in directing AI product launches
On June 12, Commerce Secretary Howard Lutnick ordered Anthropic to deny all foreign access to the Claude Fable 5 and Claude Mythos 5 models just three days after their release.
Anthropic initially announced that it would temporarily halt access to these forms for non-US citizens indefinitely, but later stated that it could not impose restrictions based on nationality and closed both forms to everyone.
The company also pushed back on claims of security concerns, calling the jailbreak process “narrow” and stating that OpenAI’s GPT-5.5 can find similar software flaws without any bypassing. Anthropic said it conducted thousands of hours of collaboration with the US government, the UK’s AI Safety Institute, and private groups before the launch.
Anthropic has since regained limited permission to sell Mythos 5 to over 100 accredited US organizations, and Getting to Myth 5 It has been restored.
OpenAI has faced similar restrictions, having to delay the full public launch of GPT-5.6 at the government’s request, limiting access to roughly 20 companies with a single government signature.
Google is also in discussions with officials before launching advanced coding models that have stronger cyber capabilities than previous generations.
The problem of state law has not disappeared
The federal crackdown was supposed to streamline the compliance process for AI companies. The December 2025 executive order specifically targeted Colorado’s algorithmic discrimination law and described state-level regulation as a barrier to innovation.
The order directed the Commerce Department to publish an assessment of the state’s AI laws within 90 days and flag those who need models to “change their truthful output.”
But a preemptive strike would require action by Congress, and that legislation has not yet been implemented. States have not given up, as some have continued to push forward with their own AI regulations.
Companies that had been pushing hard for comprehensive rules in the United States now face state-level rules they wanted to escape, and a federal apparatus that has proven willing to suspend products, issue gates, and impose export controls on short notice.
Anthropic and OpenAI are both preparing for initial public offerings. Anthropic filed a confidential S-1 filing with the Securities and Exchange Commission on June 1 after raising $65 billion at an implied valuation of $965 billion.
