A roadmap for AI, if anyone will listen

In the wake of a public conflict between the Pentagon and AI firm Anthropic, a bipartisan coalition of experts, former officials, and public figures has released The Pro-Human Declaration, a framework for responsible AI development. The document’s publication is timely, arriving as the U.S. government’s lack of a coherent AI policy was made plain by the Defense Department designating Anthropic a “supply chain risk” while OpenAI secured its own military contract. This declaration aims to fill that policy vacuum, proposing a clear path for AI that prioritizes human well-being and control.

The framework is built on five core pillars: keeping humans in charge, preventing power concentration, protecting the human experience, preserving liberty, and ensuring legal accountability for AI companies. Among its most stringent provisions are a prohibition on developing superintelligence until a scientific consensus on its safety is reached, mandatory “off-switches” for powerful systems, and a ban on architectures capable of self-replication or autonomous self-improvement. Max Tegmark, an MIT physicist who helped organize the effort, suggests starting with mandatory pre-deployment testing for AI products aimed at children, comparing the need for oversight to the FDA's role in drug safety.

The declaration's potential impact lies in its broad and unusual coalition of signatories, which includes figures as politically divergent as former Trump advisor Steve Bannon and Obama’s National Security Advisor, Susan Rice. This wide-ranging support signals a shift in the AI safety conversation from a technical debate to a mainstream political issue. By framing the stakes as a choice between a future that replaces humanity and one that expands its potential, the declaration could create the public pressure necessary to break the current legislative impasse in Washington, using concerns like child safety as a starting point for broader regulation.

While the commercial race for AI capability accelerates, the Pro-Human Declaration attempts to establish a public, safety-first baseline for regulation, leveraging the recent Pentagon-Anthropic dispute to frame the debate around national control rather than just technological progress.