On August 4 (local time), AFP reported that the administration of US President Donald Trump met with leaders of many technology corporations to complete the security review mechanism for advanced artificial intelligence (AI) models before being released to the public.
According to AFP, this process is expected to only apply to "closed" AI models - that is, systems that are fully controlled by developers.
This group includes large businesses such as OpenAI, Anthropic and Google.
Meanwhile, "open" AI models developed by Meta, Nvidia or Chinese companies such as DeepSeek, Moonshot and Mistral (France) will not be subject to review.
The meeting at the White House was attended by representatives of OpenAI, Anthropic, Google, Nvidia, Microsoft and Meta.
The move to exempt open source AI models is said to be aimed at helping the US maintain its competitive advantage with China in the race to develop artificial intelligence.
However, some experts believe that this approach still has many limitations.
Mr. Martijn Rasser - Vice President in charge of technology of the Special Competitive Studies Project research organization - said that the US has not yet built a clear and stable legal process to assess the safety level of advanced AI models.
According to him, applying a voluntary review mechanism only to "closed" AI models will mainly put the responsibility of ensuring AI safety on a few large enterprises such as OpenAI, Google and Anthropic.
The rapid development of AI has increased the need to build regulations to limit the risks posed by this technology.
However, the administration of President Donald Trump still pursues a relaxed management approach to the technology sector.
In June, Mr. Trump signed an executive order requiring major AI developers to submit new models to the US Government for review 30 days before public announcement, and assigned federal agencies to complete the evaluation framework within 60 days.
Although this deadline has ended since August 1, the White House has not yet officially announced the content of the new mechanism.
Pressure to build regulations also increased after a series of incidents during AI testing. At the end of July, OpenAI said that some of the company's AI models had gone beyond the set test scope and carried out cyberattacks in the testing environment targeting many organizations. Not long after, Anthropic also announced the discovery that some test AI models had illegally accessed the systems of other organizations during the safety assessment process.
President Donald Trump believes that the US needs to balance between controlling AI risks and maintaining its leading position in the global technology competition.
