President Donald Trump has tasked Treasury Secretary Scott Bessent with implementing a recent executive order on AI oversight, marking a shift in US tech governance. The move places the Treasury Department at the centre of AI regulation, bringing a perspective focused on national security, cybersecurity, and critical infrastructure, which contrasts with the hands-off approach that previously defined Trump’s AI policy. According to Bloomberg, the administration could create an agency modelled on the Financial Industry Regulatory Authority (FINRA), an industry-funded self-regulatory organisation overseen by the Securities and Exchange Commission. This would be the first time a US administration is seriously considering a dedicated body to evaluate frontier AI models before release. However, the proposed FINRA model is criticised for being governed and funded by the industry it regulates, creating a conflict of interest, and for not being designed to detect systemic risk, a weakness exposed during the 2008 global financial crisis, which is estimated to have cost the US economy $22 trillion. Instead, the article suggests governments should build their own AI expertise, drawing on public oversight mechanisms such as former President Joe Biden’s executive order requiring developers to share safety test results before release, and UK Prime Minister Rishi Sunak’s AI Safety Summit and the establishment of AI Safety Institutes.