Bipartisan FRONTIER Act Moves U.S. Frontier AI Oversight Toward Federal Rules
U.S. Representatives Jay Obernolte and Lori Trahan introduced the bipartisan Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting Act, or FRONTIER Act, on July 23, 2026. The bill would create a federal risk-oversight framework for advanced AI systems, defining a “frontier model” as a foundation model trained using more than 1026 integer or floating-point operations. It also proposes tiered obligations for large and very large frontier developers. A large developer would need, together with its affiliates, more than $50 million in gross revenue and at least $1 billion in AI-related research and development spending over the preceding 36 months; the proposed thresholds for a very large developer are more than $5 billion in revenue and at least $10 billion in such spending. The framework would require risk-management programs, model transparency reports, independent audits or validation, reporting of serious safety incidents and continuing evaluations. The proposal remains pending legislation and is not yet law.
Although the bill is not an intellectual property statute, its recordkeeping, audit and disclosure duties could reshape how frontier AI companies document training and fine-tuning, secure model weights, preserve testing evidence and approve technical decisions. Patent strategy may therefore need to connect earlier with regulatory compliance: companies should maintain traceable development records, separate patent-ready technical disclosures from security-sensitive trade secrets, and ensure that patent, trade-secret and compliance teams work from the same verified technical record. The final text and prospects for enactment remain uncertain, but the proposal signals a shift in the U.S. debate from broad principles toward auditable and reportable development processes.
