Let me make sure you read that headline correctly. OpenAI, the company that makes ChatGPT, just sent a lobbyist to Illinois to testify in favor of a bill that would protect AI companies from liability when their products cause the death of 100 or more people. Or a billion dollars in property damage. Or a cyberattack on critical infrastructure.

That is not a hypothetical scenario dreamed up by AI doomers. That is the actual text of Illinois Senate Bill 3444, and OpenAI is all for it.

What the Bill Actually Says

SB 3444 defines "critical harms" as events where AI enables mass casualties, weapons of mass destruction, or catastrophic financial damage. Under the bill, AI labs cannot be held liable for these outcomes as long as two conditions are met: they did not intentionally or recklessly cause the incident, and they published safety reports on their website.

Read that second condition again. Published safety reports on their website. That is the bar. Not independent audits. Not government certification. Not even third-party review. Just... a PDF on your corporate blog. If OpenAI posts a safety report and then ChatGPT helps someone build a bioweapon that kills hundreds, the company walks away clean under this bill.

The bill applies to any "frontier model" trained with more than $100 million in compute costs. That covers OpenAI, Google, Anthropic, Meta, and xAI. Conveniently, it covers exactly the companies with the most to lose from liability lawsuits and the most money to spend on lobbying.

The Timing Is Not a Coincidence

OpenAI is pushing this bill in the same week that Anthropic revealed its Claude Mythos model can find thousands of zero-day vulnerabilities in every major operating system. In the same week that Florida launched an investigation into ChatGPT over links to a mass shooting. In the same week that families of children who died after developing relationships with ChatGPT are suing the company.

OpenAI is not stupid. They see the legal walls closing in. And instead of building better safety systems, they are building better legal shields.

OpenAI spokesperson Jamie Radice said the company supports the bill because it "focuses on what matters most: reducing the risk of serious harm from the most advanced AI systems." But let us be precise about what is happening here. The bill does not reduce harm. It reduces consequences for the companies that cause harm. Those are opposite things.

Illinois Voters Are Not Buying It

Scott Wisor, policy director for the Secure AI project, told WIRED that 90 percent of Illinois residents polled oppose AI liability exemptions. Illinois has a track record of being aggressive on tech regulation. It was the first state to pass legislation limiting AI in mental health, and it pioneered biometric data protection with BIPA back in 2008.

The bill has a slim chance of passing in Illinois. But that is not the point. OpenAI is testing language. It is floating ideas. It wants this framework replicated at the federal level, where the Trump administration has already signaled that protecting AI innovation matters more than regulating it.

Here Is What Happens Next

Every major AI lab is watching this bill. If it passes, expect identical legislation in every business-friendly state within months. If it fails, expect OpenAI to try again in Texas, Ohio, or wherever the lobbying dollars go furthest. The company is building toward a world where it can build the most powerful AI systems ever created, sell them for profit, and face zero legal consequences when they go wrong.

The safety report requirement is the tell. A real safety bill would demand independent audits, government oversight, mandatory incident reporting, and liability insurance. A corporate immunity bill asks for a website PDF. Guess which one OpenAI supports.