The Evolution of Federal AI Governance: Understanding the Biden Administration’s Regulatory Framework
The Biden administration’s October 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence remains the primary federal framework governing AI in the United States. While no executive order was issued on June 2, 2026, the existing policy mandates that developers of powerful AI systems share safety test results with the Department of Commerce. This framework focuses on setting rigorous standards for cybersecurity, privacy protections, and civil rights, effectively establishing the baseline for federal AI oversight.
What is the current federal framework for AI?
The regulatory landscape is anchored by the Executive Order 14110, signed by President Biden on October 30, 2023. Unlike legislation passed by Congress, this order directs federal agencies to utilize existing authorities to manage AI risks. According to the White House Office of Science and Technology Policy, the order requires developers of models that pose a risk to national security or public safety to notify the government and submit the results of “red-teaming” safety exercises. This ensures that the federal government maintains visibility into the capabilities of the most advanced foundation models.
How does the government enforce AI safety standards?
Enforcement relies on the National Institute of Standards and Technology (NIST), which developed the AI Risk Management Framework. Agencies across the federal government use these standards to evaluate AI procurement and deployment. The Department of Commerce, through its authority under the Defense Production Act, requires companies to report information regarding the development of high-end AI clusters. By tying these requirements to infrastructure and hardware, the administration targets the physical limitations of AI scaling, making it difficult for developers to bypass safety reporting requirements.
What are the primary differences between federal and state AI regulation?
Federal policy currently emphasizes risk management and voluntary industry cooperation, whereas states have moved toward prescriptive legislative mandates. The following table highlights the distinct approaches:
| Level of Government | Primary Strategy | Focus Area |
|---|---|---|
| Federal | Executive Orders & NIST Standards | National security, safety testing, and procurement |
| State (e.g., California, Colorado) | Statutory Regulation | Consumer privacy, algorithmic bias, and liability |
While federal efforts prioritize a unified approach to prevent fragmented technological development, states like California and Colorado have pursued laws that impose direct legal liabilities on developers for harms caused by their models. According to the National Conference of State Legislatures, this state-level activity is increasing as Congress continues to debate comprehensive AI legislation.
What happens next for AI policy?
The future of AI governance hinges on whether Congress codifies these executive actions into permanent law. Current efforts, such as the SAFE Innovation Framework led by Senate leadership, seek to balance innovation with guardrails. Analysts at the Brookings Institution note that the transition from executive-led oversight to congressional legislation is necessary to provide long-term regulatory certainty for businesses. Until then, agencies will continue to refine their internal AI policies based on the October 2023 directives, focusing on the integration of AI into critical infrastructure and federal procurement processes.
Key Takeaways
- Executive Order 14110 remains the governing authority for federal AI policy, not any order from June 2026.
- Reporting Requirements: Developers of advanced AI models must share safety test data with the Department of Commerce.
- Standardization: NIST serves as the technical backbone for federal AI risk assessment.
- Legislative Gap: Federal regulation remains executive-driven, while individual states are moving faster to pass binding legislation on algorithmic accountability.