The Democratic Deficit of Contractual Governance
When private AI firms like Anthropic use contractual provisions to restrict government use of their models—such as prohibiting mass domestic surveillance or autonomous lethal weapons—they are effectively making public policy. Unlike elected officials, these firms hold no constitutional mandate and are not accountable to the public. This "reverse regulation" allows private boards to unilaterally determine the boundaries of state power, creating a system where fundamental civil liberties depend on the shifting policy preferences of corporate leadership rather than democratic consensus.
The Brittleness of Private Red Lines
Contractual protections are inherently unstable and insufficient for three primary reasons:
- Lack of Durability: Corporate policies are subject to change based on leadership shifts, competitive pressures, or fiduciary duties to shareholders. As Anthropic’s own history with its "Responsible Scaling Policy" demonstrates, voluntary commitments are often weakened or abandoned when they conflict with market realities.
- Enforcement Failures: Private companies lack the capacity to monitor how government agencies use their models in classified or sensitive environments. Furthermore, federal courts are historically reluctant to enforce private contracts against the executive branch in matters of national security.
- Displacement of Political Urgency: The existence of private red lines can suppress public demand for legislative action. When a company appears to be "solving" the problem of AI safety, the political pressure required to force Congress to act abates, leaving the public with a false sense of security while the underlying legal framework remains absent.
The Necessity of Public Law
Durable protection of civil liberties requires the slow, transparent, and public process of legislation. The author argues that even if a company's red lines are substantively correct, they are not the company's to draw. By bypassing the democratic process, these private interventions prevent the public from debating and deciding how AI should be wielded by the state. True accountability requires that these standards be codified into public law, where they are subject to electoral oversight and constitutional scrutiny, rather than hidden within the terms of a private service contract.