OpenAI publishes its Frontier Safety Blueprint
OpenAI asked Congress to build one federal framework on the model of state laws like California's SB 53, then use it to preempt those same state laws.
- Safety & alignment
- Government & policy
- Minor
OpenAI published a policy document proposing how the US federal government should regulate frontier AI, arguing for a national framework built on the emerging consensus in state laws — naming California’s SB 53, New York’s RAISE Act and Illinois’s SB 315 as models — enforced by a strengthened Center for AI Standards and Innovation (CAISI). Crucially, the blueprint asked Congress to have that federal framework preempt the state laws it would be modelled on, consolidating frontier-AI oversight in Washington.
CAISI’s role should be to conduct evaluations and recommend mitigations — not to approve or block deployments.
The document also flagged what it called early signs of recursive self-improvement in current systems — AI development itself being accelerated by AI — as a governance challenge existing institutions were not equipped to address, and called for sustained visibility into that trend.
Critics, including AI-policy commentator Zvi Mowshowitz, argued the proposal risked substituting a federal ceiling for a state floor: preemption without demonstrated federal capacity or enforcement teeth could leave frontier-AI oversight weaker than the state laws it replaced, particularly given CAISI’s evaluation-only role stopped short of approval power. The blueprint arrived as a wave of state legislation — several bills passed in 2025 and 2026 — had made preemption a live question in Washington, with OpenAI positioning itself as favouring one national standard over a patchwork it said would be harder for developers to navigate.