Illinois moves toward mandatory outside audits for frontier AI labs
Illinois lawmakers have passed a bill that would require major AI developers to have their safety practices audited by an independent third party, a step supporters say would create the strongest state-level check yet on frontier model companies in the United States.
The measure, SB 315, has cleared the Illinois House and is headed to Governor JB Pritzker, who said publicly on Wednesday that he plans to sign it. If that happens, Illinois would move beyond disclosure-based approaches in other states and require outside verification that labs are actually following the standards they claim to uphold.
What makes the bill different
State AI laws in places such as California and New York already require companies to provide information on guardrails and report safety incidents. Illinois’ proposal goes further by targeting a central weakness in many governance regimes: self-attestation.
Supporters of the bill argue that AI companies currently define their own safety standards and then largely assess themselves against those commitments. Under SB 315, that arrangement would change. An independent auditor would be required to determine whether a lab is adhering to its own safety practices.
That distinction matters. A system based on reporting can create transparency without necessarily creating accountability. A system based on third-party review adds an outside check, even if the standards being reviewed still originate with the companies themselves.
Why states are driving the fight
The bill’s progress also highlights a broader political reality: federal AI safety legislation remains limited, so statehouses are becoming the main battleground for near-term regulation. As generative AI tools spread and the companies behind them pursue larger commercial ambitions, lawmakers are under pressure to show that governance is keeping pace.
The source notes that OpenAI’s chief of global affairs told WIRED the company’s policy strategy is now oriented around passing a series of state laws. That is revealing in two ways. First, it shows companies understand that state regulation is no longer peripheral. Second, it suggests the industry sees a patchwork of state rules as the practical environment in which AI governance will be negotiated for now.
Who might perform the audits
According to the source, supporters broadly expect that labs could turn to large accounting and auditing firms such as Deloitte, EY, KPMG, and PwC. It is also possible, the article says, that companies could work with participants in the AI Evaluator Forum, a coalition of smaller research groups focused on model evaluation.
That detail points to an unresolved but important question: what kind of institution is best suited to audit AI safety? Traditional auditing firms bring scale and process discipline, but the technical depth required for frontier model evaluation may push the market toward specialized organizations or hybrid arrangements.
What the law would and would not do
SB 315 would not solve every AI safety debate. It would not create a national licensing system, define a universal set of technical benchmarks, or eliminate disagreements about which risks matter most. What it would do is establish a stronger enforcement structure around company-made commitments.
That could have practical consequences beyond Illinois. Large labs often prefer operational consistency, and once a major state requires a governance process, it can influence standards elsewhere. At minimum, the bill could become a template for other legislatures looking for a more assertive model than disclosure alone.
It also reframes a basic question in AI governance. The issue is no longer only whether companies publish policies or promise caution. It is whether someone independent checks if those promises are being followed. In that sense, the Illinois bill targets not just AI risk, but the credibility gap that has accompanied the industry’s self-regulatory posture.
If Governor Pritzker signs the measure as expected, Illinois will become a significant test case. The result will be watched closely by AI labs, state policymakers, auditors, and advocates on both sides of the regulation debate. For now, the core policy shift is clear: one of America’s largest states is preparing to replace AI safety claims on paper with a requirement for outside verification.
This article is based on reporting by Wired. Read the original article.
Originally published on wired.com
