A bipartisan coalition of researchers, former officials, and public figures just published a concrete framework for AI governance — as Washington sits paralyzed, unable to agree on even basic rules for the most powerful technology it has ever encountered.
The Pro-Human Declaration, finalized in early March 2026, lays out five pillars that its authors argue are non-negotiable: keep humans in charge of consequential decisions, prevent power concentration in the hands of AI companies or governments, protect the human experience from manipulation, preserve individual liberty, and hold AI developers legally accountable. It was signed by hundreds of experts before the Pentagon-Anthropic standoff erupted — and landed with considerably more force because of it.
The Timing That Made Everyone Pay Attention
The declaration was not written as a response to the Pentagon crisis. But it might as well have been.
On the last Friday of February, Defense Secretary Pete Hegseth designated Anthropic — whose AI already runs on classified military platforms — a "supply chain risk," a label historically reserved for firms with ties to adversaries. The reason: Anthropic refused to hand over unlimited use rights to its technology. Within hours, OpenAI stepped in and signed its own deal with the Defense Department. OpenAI's head of hardware, Caitlin Kalinowski, then resigned the following Saturday, saying the agreement lacked pre-defined guardrails on surveillance and lethal autonomy.
What emerged in the span of one week was the starkest illustration yet of how little infrastructure America has for governing AI. Congress has passed no major AI legislation. The executive branch is improvising. And two of the most prominent AI labs just ended up on opposite sides of a national security standoff with no rulebook in sight.
"This is the first conversation we have had as a country about control over AI systems," Dean Ball, a senior fellow at the Foundation for American Innovation, told The New York Times.
What the Declaration Actually Says
The Pro-Human Declaration is not academic hedging. It is specific.
Its sharpest provisions include: an outright ban on superintelligence development until there is scientific consensus it can be done safely and with genuine democratic buy-in; mandatory off-switches on all powerful AI systems; and a prohibition on architectures capable of self-replication, autonomous self-improvement, or resistance to shutdown.
On the consumer side, it calls for mandatory pre-deployment testing of AI products — particularly chatbots and companion apps aimed at younger users — covering risks including increased suicidal ideation, exacerbation of mental health conditions, and emotional manipulation.
Max Tegmark, the MIT physicist who helped organize the effort, put it in terms stripped of jargon: "You never have to worry that some drug company is going to release a drug that causes massive harm before people figure out how to make it safe, because the FDA won't allow it. Why is AI different?"
His answer: it should not be. The declaration calls for a similar pre-clearance regime for powerful AI systems before public deployment.
The Polling That Changed the Math
The political case for AI governance has historically been weak. It polls well in the abstract but lacks the visceral urgency that drives actual legislation.
Tegmark cites data showing that 95% of Americans now oppose an "unregulated race to superintelligence." If accurate, the political environment has shifted faster than the legislative calendar.
The declaration's authors are betting that child safety is the issue most likely to force a break in the congressional deadlock. Framing AI risk around children — manipulation, mental health, suicide — creates bipartisan pressure that abstract arguments about existential risk have consistently failed to produce.
What Most People Are Missing
The Pro-Human Declaration is being covered as an "AI safety" document. That framing undersells it.
Its core argument is about power concentration. The opening frames two futures: one where AI replaces humans as workers and decision-makers, accruing control to "unaccountable institutions and their machines" — and another where AI amplifies human capability without supplanting it.
Most AI governance debates in Washington focus on bias, misinformation, or cybersecurity — concrete harms that map onto existing regulatory frameworks. The Pro-Human Declaration is asking a harder question: who ends up in charge? Given what just played out between Anthropic, OpenAI, and the Pentagon, that question is no longer hypothetical.
What Happens Next
The declaration has no enforcement mechanism. It cannot compel any lab to do anything. What it does is articulate, specifically, what a set of serious people think the rules should look like — at a moment when the absence of rules just cost Anthropic its Pentagon contract and OpenAI its hardware chief.
Tegmark's approach: build public consensus first, convert it into political pressure. The child safety angle gives them a vehicle. The Pentagon drama gives them a news hook. And 95% opposition to unregulated superintelligence gives them something rare in AI policy: a genuine majority.
In Washington, "more than nothing" is sometimes where everything starts.
FOLLOW NeuralWire on X for daily AI signal — what matters, why it matters, what to do about it. →