Trump's AI Safety Plan Relies on Tech Companies to Police Themselves

The Trump administration has unveiled a plan to address risks associated with artificial intelligence that depends heavily on voluntary commitments from technology companies rather than binding regulation, according to a report from Ars Technica.

Under the approach, dozens of AI firms have agreed to submit to voluntary safety tests, a step the administration is framing as evidence of progress on AI oversight. The plan does not appear to include enforceable mandates requiring companies to comply with the testing regime or to report results publicly.

Industry self-policing

Critics of such frameworks have long argued that voluntary programs lack the teeth needed to meaningfully change behavior at large technology companies, particularly when financial incentives to deploy new AI systems quickly can outweigh caution around safety testing.

Supporters of the lighter-touch approach argue that formal regulation could slow innovation and that major AI developers have incentives of their own, including reputational risk, to take safety seriously.

The reliance on the companies themselves to design and administer safety tests has drawn scrutiny from policy observers, who note that similar self-regulatory arrangements in other industries have had mixed records of success.

It remains unclear how compliance with the voluntary tests will be verified, or what consequences, if any, companies would face for failing to meet the standards they agree to.