The Illinois House passed SB 315 by a vote of 110-0 on May 27, sending Governor JB Pritzker a frontier AI safety bill he has already pledged on X to sign. The Senate cleared it 52-5 the week before. When it takes effect on Jan. 1, 2027, Illinois becomes the first state in the country to require annual independent third-party safety audits of frontier AI labs, and the third state, after California and New York, to regulate frontier models at all.

Sponsored by Sen. Mary Edly-Allen and Rep. Daniel Didech, both Democrats, SB 315 obliges labs to publish an annual safety framework, file pre-deployment risk reports, and submit to outside audits. It carries whistleblower protections and civil penalties, and defines “catastrophic risk” to include models capable of mass harm or of causing more than $1 billion in damages through cyberattacks or loss of human control, per IAPP.

The lineup is what makes this notable. OpenAI and Anthropic both backed the bill; Anthropic’s Cesar Fernandez said it “helps establish a baseline that every leading AI developer is expected to meet.” Opposing it: Chamber of Progress, whose partners include Google, Apple, and Andreessen Horowitz, with CEO Adam Kovacevich calling SB 315 “all liability and no standards.” TechNet opposed in committee. The White House has argued comparable state provisions risk fragmenting compliance for U.S. firms.

That federal posture is the context. Days before the Illinois vote, President Trump declined to sign a planned executive order that would’ve established a voluntary federal safety-testing framework, per NBC News. The same week, OpenAI published its Frontier Governance Framework, explicitly aligning with California’s Transparency in Frontier AI Act and the EU AI Act’s GPAI Code of Practice. With Washington stepping back, Sacramento, Albany, and now Springfield are writing the rules the labs will actually answer to.

Attention now shifts to the audit market itself. Scott Wisor of the Secure AI Project told WIRED that labs are expected to tap the Big Four or smaller firms within the AI Evaluator Forum to conduct reviews, a compliance industry being summoned into existence by statute, the same way Sarbanes-Oxley conjured one in 2002.

Sources

Sources