Abdul El-Sayed is calling for a pause in training and releasing the most powerful AI systems. The Democratic Senate candidate in Michigan published his AI Under Control platform on October 13.
Independent scientists would have to certify safety before the pause ends. El-Sayed also wants to ban superintelligence and recursive self-improvement, in which AI takes over its own further development.
An independent Cabinet-level agency would inspect labs, block releases and order shutdowns. Developers would have to report unauthorized access, models escaping test environments or deception of testers within 24 hours and preserve full logs.
The plan calls for shutdown drills, external audits and whistleblower protection. It expands El-Sayed’s June platform and builds on Bernie Sanders and Greg Casar’s proposal. These are campaign proposals; a New York Times interview confirms his position.