The US and tech giants agreed on common rules for artificial intelligence control
US President Donald Trump and leaders of the largest American technology companies signed a voluntary agreement on control over the development of the most powerful artificial intelligence systems. The document provides for internal risk checks, independent audit and additional control by the companies themselves, but has no force of law.
The agreement was announced after Trump and House Speaker Mike Johnson met with technology industry leaders at the White House on September 29. Participants included executives and representatives of Meta, Google, Nvidia, Anthropic, OpenAI, Microsoft and Elon Musk's companies.
According to Axios, the document establishes common principles for companies that train and launch advanced AI models. The primary responsibility for the safety of the technologies remains with the developers themselves. Trump called the agreement "morally binding".
Meta CEO Mark Zuckerberg explained that the companies agreed on a multi-level control system. It includes internal risk assessment and protective mechanisms, checks by specialized teams, audit with the involvement of external organizations, and subsequent oversight at the level of the companies' boards of directors. Participants are also to meet regularly to develop common safety standards and practices.
The document itself allows that over time some of these requirements may be enshrined in US legislation or regulatory norms. For now, compliance with the agreements remains voluntary and no legal sanctions are provided for their non-compliance.
Such an approach aligns with the current policy of the Trump administration, which does not yet support the introduction of large-scale separate AI regulation at the federal level. The president stated after the meeting that he sees significant potential precisely in self-control by technology companies. At the same time, in Congress there remain supporters of mandatory testing of the most powerful models, disclosure of assessment results, and reporting on serious incidents.
In parallel, the US administration changed official terminology: Trump signed an executive order under which federal agencies are to use the name "Super Intelligence" instead of "Artificial Intelligence". This decision applies to new government documents and communications, but does not require rewriting already adopted regulations.
Trump also reported that the creation of a separate committee of about ten people to monitor the industry's development is being discussed. In addition, in the near future the administration plans to decide on the candidacy of a new AI policy coordinator.
Based on materials from: Axios, CBS News, Truth Social
