What the Bill Governs
On July 6 US local time, Illinois Governor JB Pritzker signed SB 315, the Artificial Intelligence Safety Measures Act. Its subjects are "large frontier developers"—companies with over $500 million in annual revenue that train cutting-edge models using large-scale compute. These companies are required to publicly disclose their safety and security practices, report significant AI safety incidents, maintain internal compliance procedures, and provide confidential reporting channels and whistleblower protections for employees who raise safety concerns. The most crucial provision is a US first: frontier models' safety protocols must undergo third-party audit. The bill passed with bipartisan support and takes effect January 1, 2027.
The Regulated Voted in Favor
Both OpenAI and Anthropic publicly supported the bill—a contrast with the two companies' past cautious attitude toward state-level regulation. With the fallout from Fable 5 being forced offline by federal export controls last month just past, the industry has clearly realized that rather than waiting for case-by-case administrative intervention, it's better to accept a set of clearly written rules. The White House is reportedly negotiating voluntary release standards with several leading labs, and which lands first—the state law or the federal framework—and how they mesh, is the next thing to watch.
What It Means for the Industry
After California's SB 53, state-level AI safety legislation takes another city, and Illinois pushes from "transparency requirements" all the way to "mandatory third-party audits." For the leading labs, compliance costs are manageable and long anticipated; the real change is that frontier models' safety claims shift from company self-attestation to an independently verifiable obligation.