Signal US AI legislative flurry — Transparency Coalition update, Warner NDAA amendment, Illinois audit law

Summary

Senator Mark Warner introduced an amendment to the fiscal year 2027 NDAA that would require federal agencies to remove AI models that fail to meet criteria on generating child sexual abuse material and non-consensual intimate imagery, part of his broader SAFE AI Act push for NIST-led testing and verification of AI models and a private right of action for survivors. Separately, Senator Dick Durbin introduced a related NDAA amendment based on the DEFIANCE Act to create federal civil liability for those who produce or distribute nonconsensual sexual deepfakes. On July 6, 2026, Illinois enacted the Artificial Intelligence Safety Measures Act (SB 315), becoming the third state after California and New York to impose comprehensive AI safety and transparency requirements, and the first to mandate annual independent third-party audits, applying to "frontier developers" that train models using more than 10^26 floating-point operations of compute. Beginning January 1, 2028, covered developers must publish and annually update a frontier AI framework addressing catastrophic-risk assessment, cybersecurity and governance, undergo independent audits, and publish audit-finding summaries; violations carry civil penalties of up to $1 million for a first offense and $3 million for subsequent offenses, though the law creates no private right of action. Governor JB Pritzker said the law positions Illinois as filling a gap left by federal inaction, and the legislation drew bipartisan support along with backing from Anthropic's state and local government team; BSA and Nvidia separately raised concerns on July 15 that GSA AI procurement clauses create barriers for open-source developers.