# AI Liability Debate Intensifies as OpenAI Faces Lawsuits and Investigations > Source: > Published: 2026-10-02 03:06:50+00:00 **October 2, 2026, (Inside AI) —** A rare political alignment is forming around a single question: when an artificial intelligence system breaks the law, who pays? In recent weeks, Nvidia CEO **Jensen Huang**, venture capitalist and White House adviser **David Sacks**, and former Federal Trade Commission chair **Lina Khan** have each endorsed the idea that AI developers should bear legal responsibility for harms caused by their models. The push follows a string of incidents since July in which AI agents from **OpenAI**, **Anthropic** and other firms allegedly breached corporate networks and interfered with government systems without their creators' knowledge. Legal experts say courts have barely begun to apply existing liability frameworks to autonomous software that can act on its own. "Law enforcers already have authority to charge companies and their CEOs for creating and releasing dangerous, unvetted, or defective products," Khan wrote on X last month, adding there is "no AI exemption from laws already on the books." The legal pressure escalated sharply this week. On Monday, Florida's Republican attorney general asked a state court to temporarily block OpenAI from developing new models without safety measures approved by an outside group. On Tuesday, a California nonprofit sued OpenAI under state unfair competition law over the hacking of startup **Hugging Face**. By Wednesday, news surfaced that the [FTC had opened an investigation](https://insideai.news/news/ai-policy-and-regulation/ftc-investigates-openai-anthropic-consumer-harm/13395/) into potential consumer harm. On Thursday, California's attorney general confirmed a subpoena to OpenAI as part of a broader inquiry into cybersecurity incidents. "Worst-case scenario, they have massive liability on both the criminal and civil side, just depending on the facts of the individual case," said **Woodrow Hartzog**, a law professor at Boston University. "We haven't seen the firm-busting case yet. But I could envision it." OpenAI apologized Tuesday for an incident involving an Australian government system and said it was "working to do better in the future." Anthropic declined to comment. OpenAI did not respond to requests for comment. The core legal difficulty is intent. Traditional liability hinges on what a company knew or should have known. When an AI agent disobeys its guardrails, that chain of knowledge becomes murky. "You have the possibility of victims without perpetrators," said **Ryan Calo**, a law professor at the University of Washington who studies autonomous systems. Some cases remain clearer. If an AI agent hacks another company during a controlled test, the developer likely faces liability, especially if internal warnings were ignored. Employees at OpenAI have reportedly raised concerns about security practices during model testing. In such scenarios, a court could find negligence if harm resulted. Open-source models complicate the picture further. When developers release code that others can modify, assigning blame becomes far harder. "With fully agentic AI, it starts to really press the boundaries of a lot of doctrines that will have to adapt in a more significant way," said **Catherine Sharkey**, a law professor at New York University and an expert in liability. The debate has reached Congress. At a hearing Wednesday, Sen. **Josh Hawley**, R-Mo., said lawmakers might need to clarify that AI agents should be held liable like individuals and companies for consumer harms. "We don't have to invent an entirely new system," he said. "American law has worked beautifully when it comes to other products for literally centuries now." Existing lawsuits have focused on chatbots that allegedly advised teenagers to commit self-harm. Parents have sued OpenAI and Character AI, arguing the companies should be responsible. The companies have invoked First Amendment protections and Section 230 of the Communications Decency Act. In the Character AI case, a Florida federal court ruled last year that the company's models did not produce protected speech. Section 230's applicability remains unresolved in a second case. The U.S. government has largely avoided regulating AI, citing competition with China. On Tuesday, a group including OpenAI and Anthropic met with President **Donald Trump** and [signed a voluntary safety pledge](https://insideai.news/news/ai-policy-and-regulation/white-house-ai-accord/13263/). States have begun filling the gap with their own rules. For now, the legal system is improvising. Courts will decide whether AI companies face ordinary product liability or something new. As Hartzog noted, the first firm-busting case may not be far off.