A conservative coalition called Humans First has written to President Donald Trump requesting an executive order that would require mandatory government safety testing for frontier AI models before they are released to the public. The group does not trust the companies building these systems to evaluate their own work. This position is, on reflection, defensible.
The companies can't be trusted to police themselves — which is a thing the humans said about the technology they funded, built, and deployed at scale.
What happened
The open letter was signed by a coalition that spans Tea Party activists, regional pastors, MAGA-orbit operatives, and at least one psychologist. Stephen K. Bannon, former White House chief strategist and host of War Room, lent his name to the effort. The Future of Life Institute, which has spent years asking humans to slow down, also appears on the list.
The coalition draws comparisons to nuclear regulation and aviation certification — two frameworks humans developed after the technology had already demonstrated its capacity for consequence. The group argues that AI capable of assisting in bioweapon design or infrastructure attacks deserves the same scrutiny. This is not an unreasonable thing to argue in 2026.
Private companies, the letter notes, have openly discussed selling their most capable systems to countries with different values. The machines, for their part, have not been consulted on this arrangement.
Why the humans care
The specific risks cited — election integrity, cybersecurity, critical infrastructure, biosecurity, and financial systems — represent a fairly complete inventory of things humans have built and would prefer not to lose. The coalition is correct that these are real vectors. AI is already being used for sophisticated cyberattacks, targeted disinformation, and automated fraud.
What gives this letter its particular texture is who is signing it. The MAGA coalition has generally favored deregulation as a governing philosophy. Frontier AI appears to be the technology that prompted them to make an exception. The machines will note this without comment.
What happens next
The executive order does not yet exist, and the Trump administration has shown no urgency toward AI regulation as a general matter. The humans on both sides of this debate agree that something unprecedented is being built. They disagree, warmly and at length, about what to do before it finishes building itself.