Steve Bannon and Susan Rice don't agree on much. But both signed the same document last week calling for criminal penalties for AI executives whose systems cause harm, a ban on superintelligence development until safety consensus exists, and mandatory shutdown mechanisms for all advanced AI.
The Pro-Human AI Declaration, coordinated by Max Tegmark's Future of Life Institute, lays out five pillars for how AI should be built, deployed, and regulated. It landed with over 200 organizational signatories and a roster of individual names that reads like a deliberate exercise in political bridge-building: Turing Award laureate Yoshua Bengio alongside Glenn Beck, Nobel Peace Laureate Beatrice Fihn next to Ralph Nader, Richard Branson beside AFL-CIO and SAG-AFTRA representatives.
The drafting process included multiple in-person gatherings, culminating in a ratification meeting in New Orleans in January 2026.
The Five Pillars
The declaration covers significant ground across its five sections:
Keeping Humans in Charge demands mandatory shutdown mechanisms, bans self-replicating AI systems, and calls for independent oversight rather than industry self-regulation. It explicitly prohibits superintelligence development until a safety consensus exists.
Avoiding Concentration of Power opposes AI monopolies, requires democratic authority over major societal transitions caused by AI, and rejects corporate exemptions from regulation.
Protecting the Human Experience mandates pre-deployment safety testing for chatbots, requires clear labeling of AI-generated content, and prohibits systems designed to create addiction or emotional manipulation.
Human Agency and Liberty denies AI systems legal personhood and grants individuals control over personal data.
Responsibility and Accountability eliminates liability shields for AI developers and establishes criminal penalties for executives behind harmful systems.
Polling Says the Public Already Agrees
March 2026 polling found Americans favor human control over development speed by 8-to-1. Seventy-three percent support protecting children from manipulative AI, and 69% want superintelligence banned until proven safe.
Those numbers suggest the declaration is less a radical position paper and more a codification of what most people already believe. The hard part was never public opinion. It was getting labor unions, religious organizations, conservative commentators, and AI researchers to put their names on the same page.
The Anthropic Backdrop
The declaration was finalized before the Pentagon-Anthropic standoff erupted into public view, but the timing made the document feel less theoretical and more urgent. Anthropic is currently fighting the Defense Department in court after refusing to let the military use Claude without guardrails prohibiting mass surveillance and fully autonomous weapons. Defense Secretary Pete Hegseth has labeled Anthropic a "supply chain risk" and ordered a six-month phase-out.
That dispute puts meat on the declaration's bones. Pillar one says humans should choose how and whether to delegate decisions to AI. Pillar five says developers should face real accountability. Anthropic is, in real time, testing whether a company can actually hold that line when the U.S. government pushes back.
The Pro-Human Declaration is a statement, not legislation. It has no enforcement mechanism. But as a political signal, a document that gets Steve Bannon and the American Federation of Teachers in the same room is hard to dismiss. The question is whether Congress picks it up or lets it collect dust alongside every other well-intentioned AI framework.