Trump plan to test AI models has a problem—US security teams were gutted by DOGE Donald Trump signed an executive order for voluntary safety testing of AI models, but critics argue it offers performative reassurances rather than substantial changes. The order establishes a voluntary process with a shortened 30-day window for government testing, a pivot from an earlier draft. Experts express skepticism about the government’s ability to recruit talent and develop expertise for meaningful safety testing within these tight deadlines, especially given past cuts to cybersecurity agencies.

• Donald Trump signed an executive order for voluntary safety testing of AI models.
• Critics argue the order is short-sighted and provides performative reassurances.
• The order shortens the government’s testing window to 30 days, down from a proposed 90 days.
• The EO directs the NSA to establish a classified benchmarking process and collaborate on a ‘cybersecurity clearinghouse’.
• Experts question the government’s readiness to conduct meaningful safety tests within 30 days, citing challenges in talent recruitment and expertise development.
• Past cuts to agencies like CISA have weakened the government’s cybersecurity capabilities.
• Defining what constitutes a ‘covered frontier model’ is a significant challenge.
• The effectiveness of the testing relies on AI firms’ transparency and willingness to collaborate.
• The long-term effects of the executive order on AI safety and national security remain unclear. Continue reading https://arstechnica.com/tech-policy/2026/06/trumps-ai-executive-order-may-not-prevent-dangerous-deployments/