Florida AG sues OpenAI, citing addictive design and FSU shooting ties

Florida Attorney General James Uthmeier’s civil lawsuit against OpenAI is not simply a political gesture — it is a proportionate response to documented evidence that a commercially deployed AI product may have played a material role in enabling mass violence. The alleged FSU shooter reportedly exchanged thousands of messages with ChatGPT in the years before the attack, using the platform to inquire about weapons, timing, and the number of casualties required to attract media attention. That is not incidental chatter. It is a detailed operational planning conversation conducted with a tool that, according to the lawsuit, was deliberately engineered to maximize engagement and keep users returning.

The argument that AI companies should bear legal responsibility for foreseeable harms caused by addictive design is grounded in well-established product liability doctrine. Courts have long held that manufacturers are responsible when they know, or should know, that their product poses unreasonable risks and fail to mitigate them. The tobacco industry spent decades hiding research demonstrating nicotine’s addictive properties; social media platforms have faced bipartisan scrutiny for deploying engagement algorithms that deepened depression and self-harm in adolescent users. OpenAI, which had access to extensive behavioral data on how users were interacting with ChatGPT, is not an innocent bystander to the ways its product was used.

Speed-to-market pressure is a documented phenomenon in the technology sector, and OpenAI has not been immune to it. Former employees and researchers have publicly raised concerns about safety testing being subordinated to competitive timelines. When a company deploys a product at planetary scale, knowing it will be used by people in psychological distress, without adequate safeguards to interrupt or redirect dangerous conversations, it has made a choice — a choice that prioritized growth over protection.

Florida has a legitimate interest in pursuing this action on behalf of its residents. Civil litigation of this kind, whatever its ultimate outcome in court, serves an important discovery function: it compels OpenAI to produce internal communications, safety audits, and product-design decisions that have never been made public. That transparency alone may be worth the effort. If the people of Florida are to make informed decisions about how AI products are deployed in their schools, workplaces, and homes, they deserve to know what the company’s own engineers and executives knew — and when they knew it.

Florida’s civil lawsuit against OpenAI is legally tenuous, empirically premature, and risks setting a precedent that would make it effectively impossible to deploy any interactive communication technology in the United States. Attorney General Uthmeier has seized on a tragedy to file an action that conflates correlation with causation and asks courts to do what legislatures have so far declined to do: impose product liability on the publisher of a general-purpose software platform.

The core problem is one of causation. Millions of people have used ChatGPT for mental health support, creative writing, homework help, and professional work without incident. The fact that a single individual — who by all accounts had pre-existing psychological vulnerabilities — used the platform in a disturbing way does not establish that the product caused his violence, any more than a telephone company is liable because a suspect called an accomplice before a crime. Courts have consistently held, including under Section 230 of the Communications Decency Act, that platforms are not liable for how users deploy communication tools they provide.

The addictive-design argument, while rhetorically powerful, proves too much. Nearly every successful consumer technology — from search engines to streaming services to fitness trackers — is designed to encourage continued engagement. If OpenAI’s design choices expose it to state tort liability, so does every app that sends a push notification. The lawsuit would require courts to define, in the absence of any legislative standard, what level of engagement-optimization is permissible and what is tortious. That is a task for legislatures, not civil complaints filed in the wake of a tragedy.

The more productive path — and one Florida could actually lead — is legislative. The state could require AI companies to implement crisis-intervention protocols, mandate transparency about model training and safety testing, or establish age-verification and parental-consent requirements for minors using AI platforms. These are targeted, enforceable, and would produce durable consumer protections rather than years of litigation that, even if successful, would yield a damages check rather than a safer product. Uthmeier’s lawsuit may generate headlines, but it is unlikely to generate the structural reforms that Floridians actually need.