US cross‑ideological coalition demands strict human control over AI

The most interesting part of this story isn’t that another AI principles document exists; it’s who is signing it. Getting Steve Bannon‑aligned conservatives, progressive activists, unions and religious leaders to co‑sign anything is rare in U.S. politics, and they are uniting around a very specific idea: highly autonomous AI must remain under “meaningful human control” with genuine, independent oversight, not just self‑regulation by labs. That’s a direct challenge to the implicit assumption inside many frontier companies that deployment speed is the overriding objective and that safety can be managed largely in‑house.([nampa.org](https://www.nampa.org/text/22877320))

For the race to AGI, this coalition is a signal that the political Overton window is moving. As models become more agentic—planning, executing multi‑step actions, operating physical systems—the demand for external veto points and pre‑deployment review will grow louder. If this coalition can translate its declaration into concrete legislation or procurement rules, it could slow down certain classes of risky autonomous systems even as corporate and geopolitical incentives push the other way. The fact that the same document explicitly rejects “corporate welfare” for AI labs suggests future public funding and access to critical infrastructure could be conditioned on compliance with fairly strict control norms.

In practice, this won’t stop labs from pushing capabilities, but it may shape the runway: more scrutiny on defense and surveillance contracts, more pressure for third‑party audits, and tighter expectations around human‑in‑the‑loop for safety‑critical uses. That could modestly delay the most dangerous forms of unbounded agentic AI while still allowing productivity‑oriented systems to proliferate.