The Donald Trump administration is reconsidering its long-held free-market approach to artificial intelligence, now putting state oversight on its agenda. According to US officials, the White House is evaluating the creation of a review and oversight mechanism for newly developed AI models.
According to a New York Times report, one of the options the White House is considering is establishing an AI working group that would bring together technology companies and government officials. This group is planned to examine how oversight mechanisms would function. Among the possibilities discussed is the creation of a formal review process for newly developed AI models.
Anthropic, Google, and OpenAI Executives Consulted
Last week, White House officials met with executives from Anthropic, Google, and OpenAI to share some details of these plans. During the discussions, it was stated that a model similar to the system implemented in the UK, which monitors whether AI models meet safety standards, could also be applied in the US.
This development signals a significant shift in the Trump administration's approach to artificial intelligence. Since returning to office, Trump has viewed AI as a critical element in competition with China and had largely relaxed regulations in the sector. Notably, the previous administration's regulations, which included safety tests and reporting requirements, had been abolished.
Trump had previously emphasized that AI should not be hindered, arguing that excessive regulation could stifle innovation. However, recent developments indicate that the administration is seeking a more balanced approach.
Anthropic's Mythos May Have Been Influential
This policy shift is attributed to Anthropic's new AI model called "Mythos." The company stated that this model is highly effective at detecting software vulnerabilities and could pose serious cyber risks if misused, leading them to choose not to release it publicly.
The White House is reportedly seeking to avoid the political consequences of potential large-scale AI-powered cyberattacks, while also evaluating the technology's potential in defense and intelligence fields. In this regard, some officials are proposing an oversight system that would provide the government with early access to new AI models but would not prevent their publication.
Warning: Could Hinder Competition with China
The new approach has elicited mixed reactions in the technology world. Some company executives argue that excessive government intervention could weaken the US's competitive edge against China, while there is no clear consensus within the industry on how regulation should be implemented.
Dean Ball, who previously served as an AI advisor in the Trump administration, highlighted the difficulty of the process, stating, "Technology is advancing very quickly, but excessive regulation is also not desired. Striking this balance is quite challenging."
Changes in Administration and Crises
The shift in AI policy coincides with changes in personnel at the White House. David Sacks, who was in charge of AI policies, left his post in March, while White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent became more actively involved in the process.
Furthermore, the dispute between the Pentagon and Anthropic complicated the process. Following a disagreement over a $200 million contract and the military use of AI, the Pentagon ceased using the company's technology. Anthropic, in turn, filed a lawsuit against this decision.
Despite this, the company's AI systems continue to be used in the military analysis platform called "Maven." Additionally, the US National Security Agency is also reportedly benefiting from the Mythos model.
Officials state that it is not yet clear which institutions will be involved in a potential oversight system. However, it is suggested that the National Security Agency, the Office of the National Cyber Director, and the Office of the Director of National Intelligence could play a role in this process.
It is also being discussed that the Artificial Intelligence Standards and Innovation Center, established during the previous administration, could be reactivated in this process. This institution was created to evaluate voluntarily shared AI models but had fallen into the background recently.
0 Comments: