Rethinking at the White House: Trump administration considers more AI oversight

The White House has largely left domestic AI development to its devices so far. A course that is now apparently being reevaluated: The Trump administration is considering the establishment of an AI task force, consisting of tech leaders and government representatives, to develop possible oversight mechanisms for artificial intelligence, the New York Times reports. The newspaper refers to unnamed US officials and individuals familiar with the discussions. According to them, White House representatives had already discussed corresponding plans with executives from Anthropic, Google, and OpenAI last week.

Among the measures discussed is a formal government review process for new AI models, which would give the government early access to the systems without blocking their release. Should the Trump administration actually introduce a pre-approval process, the task force would also have to determine which authorities would be involved, the report states.

The NSA, the White House Cyber Coordination Office, and the Director of National Intelligence (DNI) could play a key role. In addition, the task force could clarify whether the “Center for AI Standards and Innovation” created by the Biden administration, which has largely been sidelined under Trump, will be involved.

Mythos shakes up Trump's AI course

So far, the White House's AI policy has focused on deregulation and rapid expansion of the domestic AI industry to avoid falling behind in the AI race with China. Among the first steps of the Trump administration was the reversal of a Biden regulation that required AI companies to conduct security tests for particularly powerful models and report security-relevant information to the government. Ironically, it is precisely in the area of cybersecurity that this course is now apparently reaching its limits. According to sources from the New York Times, the recent rethinking is said to be due to concerns about the political consequences of a devastating, AI-powered cyberattack.

At the beginning of April, Anthropic introduced the cybersecurity model Mythos, which is particularly effective at finding and exploiting security vulnerabilities and was therefore only made accessible to selected partners, including US government agencies and security authorities. These were intended to find and fix vulnerabilities in critical software. According to the New York Times, the NSA (National Security Agency) also had advance access to Mythos. The security policy significance of Mythos likely contributed to the Trump administration again seeking dialogue with the company, despite the ongoing conflict between Anthropic and the Pentagon.

The White House described reports of possible government AI oversight as speculation. Any political announcements will be made by Trump himself, it is said to the New York Times.

(afl)