 |
| Artificial intelligence/photo/Igor/Unsplash |
The United States is in advanced discussions with several leading artificial intelligence companies to establish voluntary standards governing the release of advanced AI models, according to a report published Wednesday, as the Trump administration seeks to strengthen oversight of rapidly evolving technology without imposing mandatory regulations.
An announcement outlining the framework could come as early as next week, the Financial Times reported, citing people familiar with the negotiations.The proposed standards are expected to establish voluntary benchmarks for evaluating increasingly powerful AI models before they are made widely available. They would also provide guidance on deployment timelines and clarify which users, organizations and foreign entities should be permitted access to the technology, according to the report.
The initiative reflects Washington’s growing focus on national security as generative AI systems become more capable. U.S. officials have increasingly expressed concern that cutting-edge models could be exploited by military or intelligence organizations in countries such as China and Russia if released without sufficient safeguards.
Although Reuters could not independently verify the report, the discussions are consistent with broader efforts by the administration to tighten scrutiny over advanced AI development while avoiding regulations that industry leaders argue could slow innovation.
The White House did not immediately respond to requests for comment outside normal business hours. Neither did AI developers Anthropic and OpenAI.
The negotiations follow a series of policy measures introduced by President Donald Trump aimed at expanding federal oversight of advanced artificial intelligence.In June, Trump signed an executive order directing federal agencies to collaborate with major AI developers in evaluating advanced models before public deployment. The order also instructed agencies to draft standards intended to assess potential national security and public safety risks associated with increasingly capable systems.
Unlike mandatory regulatory requirements, the proposed framework would rely on voluntary commitments from participating companies, allowing developers greater flexibility while encouraging consistent risk assessment practices across the industry.
Google is among the companies engaged in discussions with the U.S. government ahead of the planned release of new coding-focused AI models that offer more advanced programming capabilities, according to a source familiar with the matter.
The source said Google has also participated in wider conversations about establishing common industry standards, with the Financial Times first reporting those discussions.The talks come as federal authorities continue to intervene directly in the deployment of some of the world’s most advanced AI systems.
Earlier this week, the Commerce Department lifted export controls that had temporarily restricted Anthropic’s latest Fable and Mythos models. The restrictions had been imposed less than three weeks earlier after officials raised national security concerns over the models’ international availability.
The decision highlighted the government’s increasingly active role in determining how advanced AI technologies are distributed globally.
OpenAI has also experienced government intervention.Last week, the company delayed the broader release of its GPT-5.6 model at the request of U.S. officials. Instead of an immediate public launch, access was initially limited to a small group of vetted partners while additional evaluations were conducted.
The delay underscored the extent to which Washington has become directly involved in decisions surrounding the deployment of frontier AI systems, particularly those considered to possess strategically significant capabilities.
Federal officials have repeatedly argued that stronger oversight is necessary because advances in artificial intelligence are occurring at an unprecedented pace.
Modern large language models are increasingly capable of writing sophisticated software, conducting scientific analysis, generating complex technical documents and assisting with cybersecurity tasks. While those capabilities create significant commercial opportunities, policymakers also worry they could facilitate malicious cyber activity, disinformation campaigns or military applications if deployed irresponsibly.
The voluntary standards under discussion are expected to address some of those concerns by encouraging developers to perform additional safety testing before releasing their most capable systems.
Industry leaders have generally supported closer cooperation with governments while opposing rigid regulatory frameworks that they believe could disadvantage American companies competing against foreign rivals.
Many AI executives have argued that voluntary standards provide a more flexible approach, allowing safeguards to evolve alongside rapidly changing technology rather than becoming outdated through lengthy legislative processes.
The discussions also come at an important moment for the commercial AI industry.Both OpenAI and Anthropic are preparing for initial public offerings, increasing investor attention on how government oversight could affect their future growth and product development.
Clearer guidance from Washington could help reduce regulatory uncertainty for both companies while reassuring customers and investors that advanced AI models are being deployed under consistent safety expectations.
Technology companies have invested billions of dollars in developing increasingly powerful AI systems, fueling intense competition across the sector.
Major cloud providers, semiconductor manufacturers and software developers have likewise expanded investment in AI infrastructure, making the deployment of frontier models a critical commercial priority.
At the same time, governments around the world are struggling to develop policies that encourage innovation without creating unacceptable security risks.
The United States has generally favored close collaboration with private industry rather than comprehensive regulation, contrasting with approaches adopted in some other jurisdictions that rely more heavily on statutory rules. The emerging voluntary framework appears intended to reinforce that strategy by establishing shared expectations while preserving the industry’s ability to innovate rapidly.