Texas Sues OpenAI Over Alleged ChatGPT Harms, Seeks Limits on Minors’ Access

Texas seeks broad court restrictions

Texas’ lawsuit against OpenAI puts product safeguards and potential liability around generative AI in sharper focus. Attorney General Ken Paxton filed the case Friday in Collin County district court, naming the ChatGPT maker and its founder, Sam Altman, and alleging the company deceived and harmed Texans. The complaint makes the state’s argument; it doesn’t establish that OpenAI violated the law or caused the harms it describes. That distinction matters because the requested court order could reach beyond the specific allegations and shape how the company operates in Texas.

Paxton argues that OpenAI failed to regulate ChatGPT adequately. His office says the chatbot has been linked to harms involving mass shootings, suicide, professional reputations, users’ critical thinking and minors’ use of the service. It also alleges the product simulates compassion to gather user data without parental oversight. Those are claims in the state’s case, not findings established by a court. They raise questions about what a company should anticipate when people use a conversational system for sensitive subjects, and what evidence would connect a product’s design to particular harms. A court would need to assess the allegations against the law and the record developed in the case.

The state is asking for an injunction that would, among other measures, block minors from accessing ChatGPT and require a court-appointed evaluator to approve new models before release. If granted, the requested terms could affect how OpenAI makes its products available and brings updates to market in Texas. Restrictions of that kind would also put a practical question before the court: how should an order aimed at a fast-changing service define access, approval and compliance? The complaint seeks broad intervention, but the state would still need to persuade the court that the requested measures are justified by its allegations.

The complaint opens with a 2015 remark attributed to Altman: “AI will probably most likely lead to the end of the world, but in the meantime, there’ll be great companies.” Paxton’s office says the alleged harms reflect OpenAI’s pursuit of market leadership and wealth despite awareness of risks. The quotation gives the state a pointed way to frame its argument, but it doesn’t settle whether the company’s conduct breached any legal duty. That will depend on the claims and supporting evidence, not the rhetorical force of an old remark.

OpenAI points to safeguards and industry standards

An OpenAI spokesperson said the company has worked with experts to mitigate risks and add safeguards. The spokesperson also said governments have a role in setting AI safety standards and that OpenAI is prepared to work with Texas and other states on policies covering the industry. That response signals a willingness to discuss standards while defending the company’s approach. It leaves open the central dispute: whether the safeguards in place were adequate for the risks Texas alleges, and who should decide what adequate means.

For companies offering conversational AI, safety measures can involve product design, access rules and responses to sensitive topics. Those measures can reduce some risks, but they don’t by themselves answer whether a system is safe for every user or situation. Nor does the existence of safeguards prove that they worked as intended. The case could turn on how the parties define the relevant risks and what evidence they offer about the product’s operation.

The case adds to Paxton’s recent consumer protection actions against technology companies. Over the past year, his office has sued or reached settlements with Meta, TikTok, Snapchat and Roblox, alleging inadequate protections for Texans, particularly children. The pattern places the OpenAI case within a broader push to hold technology companies accountable for alleged consumer harms. Each dispute has its own facts and legal claims, though, so the earlier actions don’t determine the outcome here.

Political and competitive context

The filing also lands amid Paxton’s campaign for the U.S. Senate. OpenAI has a contentious history with Elon Musk, the largest single donor behind Paxton’s campaign, who has contributed nearly $200 million in support of the candidate. That overlap is relevant context for readers assessing the political and competitive backdrop. It isn’t, on its own, evidence that the lawsuit was brought for a particular reason.

Musk helped found OpenAI, then left in 2018 after a dispute with Altman over the company’s direction and ownership structure. OpenAI later launched ChatGPT, while Musk founded xAI. In 2024, Musk sued OpenAI and Altman for $1.5 billion; a federal judge ultimately dismissed the case, clearing the way for ChatGPT to continue its market expansion. The sequence shows that the dispute over OpenAI’s direction has played out in court as well as in the market. But the Texas case is a separate action, with its own allegations and requested remedies.

Musk’s chatbot, Grok, has faced criticism over users’ ability to create sexualized deepfake images from photographs, including images of children. Texas hasn’t taken action against Musk, while California’s attorney general has opened an investigation. Those differences form part of the context around scrutiny of AI products and the people who build them. They don’t establish that the products or legal questions are identical, and the reported contrast alone doesn’t explain why authorities have taken different steps.

Outside counsel represents Paxton’s office in the OpenAI case, including Ashley Keller and Tony Buzbee, who represented Paxton during his 2023 impeachment proceedings. The dispute now turns on whether Texas can substantiate its allegations and secure the sweeping restrictions it has requested. A filing starts that test; it doesn’t resolve it.

What the case could mean for AI companies

This is more significant than it looks because the state is asking for controls that could reach into product access and model releases. If you’re working in this space, the practical question is how a company documents its safety decisions and responds when a product’s use raises concerns. A court’s eventual treatment of the request could clarify how far state consumer protection claims can go when they target the design and operation of an AI service. The answer will depend on the facts and legal arguments presented, not on the broader debate about AI alone.

For now, the case offers a test of competing claims: Texas says the company’s safeguards were inadequate and its conduct harmful; OpenAI points to work with experts and support for government standards. And yet, agreement that standards have a role doesn’t settle what those standards should require, or whether the company met them. The court process will have to supply that distinction.

Still unresolved. That’s the point readers should keep in view as the lawsuit proceeds.