Illinois Passes the Toughest AI Law in the US: Safety Audits Become Mandatory

Illinois has become the first US state to pass a comprehensive AI safety law. Lawmakers are requiring OpenAI, Anthropic, Google and similar organizations to prove to third parties that they meet strict safety standards. The law aims to create a transparent oversight mechanism in a fast-moving industry where risks often remain unclear even to the developers themselves.

What the new law requires

The bill obliges developers of large language models and AI systems to undergo independent safety audits. The third parties are specialized firms or organizations that must objectively confirm that companies follow the established protection standards. The law defines mandatory areas of review and transparent compliance rules that go far beyond internal corporate assessments.

The law's requirements cover several key areas:

• Independent review of algorithms for bias, discrimination and unequal impact on population groups
• Testing resilience against malicious use, manipulation, social engineering and abuse
• Assessment of potential effects on the labor market, education, social inequality and society as a whole
• Documentation of safety measures, control mechanisms, risk management, reporting and accountability
• Public disclosure of audit results, vulnerabilities found and improvement measures

The law makes regular audit updates mandatory: at least once a year, or whenever the system changes significantly. This is needed to track the evolution of AI systems, new types of risk and emerging uses that may create unforeseen problems.

History and current context

Until now, the US has had no single federal AI safety law, although various proposals are under discussion in Congress. Several states have tried to adopt their own regulatory measures, but Illinois is the first to create a truly comprehensive and binding regulation, rather than recommendations or voluntary standards.

This comes amid growing pressure from activists, independent researchers, tech-company employees and politicians calling for tighter control over the development of powerful AI systems. Scandals involving discriminatory algorithms, data leaks and manipulation have shown that industry self-regulation is not enough.

Governor JB Pritzker has already formally expressed his readiness to sign the bill, which means it will take effect and begin to be put into practice. This could become an important precedent for other states and significantly influence federal AI regulation nationwide.

What this means for companies

Companies such as OpenAI and Anthropic already work on internal safety standards and regularly run self-assessments, but the law introduces mandatory external oversight by independent third parties for the first time. This creates an objective verification mechanism that is harder to circumvent or manipulate than internal reports.

For startups and smaller AI companies, this may mean extra costs for professional audits and consulting, but it will also make market entry easier. They will be able to use approved, proven standards instead of developing their own processes.

The law may complicate AI companies' operations in Illinois and raise compliance costs. In theory, it could push some players to move their headquarters to states with softer regulation. But companies that have already invested in safety, transparency and ethics will gain a competitive advantage: they will be able to pass audits and obtain certification faster.

Global impact

Illinois shows that US states are ready to regulate AI on their own in the absence of a reliable federal law. This precedent could speed up similar laws in other US regions and countries, gradually shaping de facto global safety standards.

The European Union is already developing its AI Act with documentation and transparency requirements, but the Illinois law shows that regulation can be decentralized. Each region sets its own requirements and compliance standards. This creates a complex but potentially healthy landscape for global companies, which now have to adapt to different requirements at the same time.

What this means

Illinois passing this law is an important signal that the era of AI self-regulation is ending. Society, regulators and even developers themselves are demanding transparency and accountability. The law could become a model for other regions, gradually turning AI safety from a voluntary practice into a mandatory norm, as happened with other critical technologies in the past.