Paxton Sues OpenAI, Seeks ChatGPT Limits in Texas

Estimated Time to Read: 13 minutes

Texas Attorney General Ken Paxton (R) has sued OpenAI, several affiliated companies, and CEO Sam Altman, alleging that OpenAI misled Texans about ChatGPT's safety and reliability and failed to provide protections required for minors.

The lawsuit, announced October 23, invokes the Texas Deceptive Trade Practices Act (DTPA) and the Securing Children Online through Parental Empowerment Act, House Bill 18 (HB 18), commonly called the SCOPE Act. It seeks civil penalties and extensive restrictions on how OpenAI offers its products in Texas.

The most consequential requests extend beyond advertising disclosures. Texas asks the court to prohibit OpenAI from offering ChatGPT to minors in Texas and to require a court-approved independent evaluator to review and approve new artificial intelligence models before their release to Texas users.

Those requests bring a larger policy question into focus: How should Texas enforce existing consumer protections without letting litigation replace legislative debate, or letting the state replace parents' judgment about their own children's use of technology?

What the Texas OpenAI Lawsuit Alleges

The state's 67-page original petition argues that OpenAI's marketing creates expectations of safety, accuracy, and trustworthiness that ChatGPT does not consistently meet.

Texas points to allegedly inaccurate legal and medical information, dangerous responses involving self-harm and violence, and conversational features that the state contends encourage emotional dependence. The petition also alleges that OpenAI prioritized product releases and commercial growth over adequate safety testing. Those claims are framed as deceptive trade practices under the DTPA, not as violations of the SCOPE Act's content-filtering requirement, which remains blocked in federal court.

These are allegations advanced by the state. The petition does not establish that OpenAI caused every harm it describes, and some examples rely on claims made in other lawsuits or in published reporting.

The lawsuit also seeks to hold Altman personally liable, alleging that he directly participated in the company's disputed representations and decisions. Whether the evidence supports those allegations, and personal liability for Altman, is a question for the court.

ChatGPT Parental Controls and the SCOPE Act

A separate part of the lawsuit concerns how ChatGPT handles minors' accounts, personal information, and parental oversight.

Texas alleges that OpenAI fails to adequately verify a parent's identity and relationship to a minor and does not provide all the parental tools required by the SCOPE Act. The petition specifically challenges controls involving minors' privacy settings, personal information, and targeted advertising.

Portions of the SCOPE Act remain blocked in federal court. In July, the Fifth Circuit upheld an injunction against the law's content monitoring and filtering requirement, finding it preempted by Section 230 of the Communications Decency Act. The provisions at issue here, involving parental verification, parental tools, and targeted advertising to known minors, were not part of that injunction and remain in effect.

The claims also mirror the state's separate 2024 SCOPE Act lawsuit against TikTok, which involves similar requirements for minors' personal information, parental verification, and account controls and remains unresolved.

The petition acknowledges that OpenAI introduced parental controls in September 2025. The dispute concerns whether those controls satisfy Texas law, including whether voluntary account linking provides sufficient oversight.

According to The Texas Tribune, OpenAI responded that it has worked with experts to reduce risks and strengthen safeguards. The company also expressed support for AI safety policies that apply across the industry.

For policymakers, the distinction matters. Giving parents useful tools to direct their children's technology use is a different policy choice from prohibiting access regardless of parental permission. Texas Policy Research (TPR) has described those parental-empowerment provisions as the strongest portions of the SCOPE Act, and enforcing them is a different matter from the broader restrictions the state now seeks.

Texas Seeks a ChatGPT Ban for Minors and Pre-Release AI Reviews

The petition's requested injunction would reach well beyond correcting specific marketing claims.

Texas asks the court to bar OpenAI from offering ChatGPT to minors in Texas. As written, that request does not include an exception for parental consent or supervised educational use.

The state also seeks to prevent the release of any new AI model to Texas users unless a court-approved independent evaluator has reviewed its safety guardrails and signed off on its release.

If granted, that provision would establish an ongoing approval requirement affecting access for adult Texans as well as children. The proposed restriction therefore raises questions about the evaluator's standards, authority, accountability, and ability to delay access to new products.

The petition additionally requests civil penalties of up to $10,000 for each DTPA violation, disgorgement of assets as permitted by law, and reimbursement of investigation costs and attorneys' fees. These are requested remedies, not penalties or restrictions already imposed.

Section 230 and AI-Generated Content

The Fifth Circuit's July ruling relied on Section 230, which generally prevents states from treating online platforms as the publisher or speaker of content provided by others. Because the SCOPE Act's filtering requirement would have forced platforms to monitor and remove users' content according to state-defined standards, the court held it preempted by federal law.

That protection may not carry over to this case. ChatGPT's responses are not posted by other users; they are generated by OpenAI's own system. Section 230 shields platforms from liability for information provided by another party, and whether it covers output an AI model produces itself remains unsettled. Courts have not definitively resolved the question, and early cases involving conversational AI have largely proceeded on legal theories that do not depend on Section 230.

The state's core claims also concern OpenAI's own conduct rather than third-party speech. Marketing statements and account practices are the company's own representations and design choices, which Section 230 was never intended to protect.

That difference cuts both ways. Section 230 remains one of the primary legal safeguards for free expression online, and this case should not become a pretext for weakening its protections for platforms that host their users' speech. At the same time, if AI-generated output falls outside Section 230, the rules governing liability for that output should be set deliberately. As the Fifth Circuit observed, changes to the federal framework belong with Congress. How Texas law applies to AI systems is likewise a question for the Legislature, not for court-ordered remedies in a single enforcement action.

Consumer Protection and AI Accountability

Companies should be accountable for materially deceptive claims about their products. If evidence establishes that OpenAI misrepresented specific capabilities or failed to meet enforceable statutory obligations, those findings would provide a basis for appropriate relief.

But the existence of inaccurate or harmful chatbot responses does not, by itself, establish deceptive marketing. The court will need to assess the representations and practices at issue under the applicable legal standards.

Personal responsibility also remains relevant. People using AI for consequential decisions must evaluate its outputs, and professionals remain responsible for work they submit or advice they provide. That responsibility can coexist with a company's obligation to accurately describe its product.

A sound approach should preserve both forms of accountability. Consumers need truthful information about limitations, and companies need clear legal standards against which their conduct will be judged.

AI Age Verification and Digital Privacy

The lawsuit also places age verification at the center of the policy debate.

Distinguishing minors from adults and verifying parent-child relationships can create pressure to collect additional personal information. Depending on implementation, those requirements could affect adults seeking lawful access as well as families using parental controls.

The petition does not specifically ask the court to require every Texan to submit government-issued ID. Nevertheless, policymakers should examine what information compliance would require, how it would be protected, and whether less intrusive methods could accomplish the same objective.

The state's case against Discord shows how these questions are already being decided through litigation. In July, Texas secured an agreed temporary injunction requiring Discord to extend the age assurance measures and default safety settings it uses in the United Kingdom to Texas users within 90 days. That order does not resolve whether Discord violated the DTPA, but it establishes platform-wide requirements without legislative debate over their scope or privacy implications. The state's own petition against Discord raised concerns about how the company protected identification documents users had submitted for age verification, a reminder that age verification creates its own risks.

Discord is not the only example. In September, a Travis County court found that TikTok violated the DTPA by misrepresenting its content moderation policies and Restricted Mode feature, and the state has negotiated a settlement requiring Meta to change teen screen-time limits, parental controls, and age-assurance practices. Each action rests on different legal grounds, but together they show Texas increasingly shaping platform design, age assurance, and default settings through litigation.

The TikTok ruling also points to a narrower path. The state established deception by examining the company's own representations and internal practices, without requiring any Texan to prove their identity first.

Age verification also rarely stays confined to one platform. When 27 attorneys general backed the App Store Accountability Act at the Supreme Court, we warned that lawmakers could extend similar requirements to social media, gaming platforms, and artificial intelligence tools. This lawsuit shows that the extension can come through litigation as easily as legislation. Each new requirement moves Texans closer to a de facto digital ID system, in which ordinary online activity depends on a verified real-world identity.

That creates real security risks. In June, a cybersecurity incident at a Texas Parks and Wildlife Department (TPWD) license vendor affected more than 3 million hunting and fishing license customers. Every database of sensitive personal information becomes a potential target, and a company cannot lose information it never collected.

Protecting children's privacy should include scrutiny of any new information collection prompted by government requirements, and that scrutiny should extend to the adults whose information would be collected as well.

Child Protection Should Not Replace Parental Responsibility

No serious person minimizes the risks children can face online. The state has a proper role in prosecuting predators and in enforcing consumer protection laws when companies deceive parents about the safety of their products. But not every remedy advanced under the banner of child protection deserves automatic support.

The requested ban on minors' access illustrates the problem. As written, it would decide for every Texas family that no minor may use ChatGPT, whether a parent wants a teenager to use it for schoolwork under supervision or prefers to keep it out of the house entirely. That decision belongs to parents, not to the attorney general or a court.

The contrast with other Texas policy is telling. Texas Policy Research opposed the App Store Accountability Act because it inserts a state-designed verification and consent structure into decisions that belong to families. The relief sought against OpenAI would go further still. Rather than conditioning a minor's access on parental consent, it would remove parents from the decision altogether.

This case also fits a broader pattern. From proposed social media bans for minors to the federal Kids Online Safety Act (KOSA) and the debate over school phone policies, policies framed as child protection increasingly shift authority from families to the state. Each step may appear reasonable on its own. Over time, they move the line between protecting children from genuine harm and substituting government judgment for parental responsibility.

The better framework empowers parents rather than replacing them. That means truthful representations from companies, parental tools that are transparent and easy to use, and enforcement against deception. It does not mean accepting that the state should become the default guardian of every child's digital life. Texas should not become the digital parent.

Texas AI Policy and Legislative Authority

This lawsuit is not an example of the attorney general acting without any claimed statutory authority. The petition invokes enforcement powers under the DTPA and the SCOPE Act. The more difficult question concerns the scope of relief available under those laws and whether the requested restrictions are justified by the violations the state can prove.

Courts have a legitimate role in remedying unlawful conduct. However, a prohibition on minors' access and an ongoing approval requirement for new AI models would have substantial consequences for parental choice, free enterprise, and access to digital tools.

Those consequences deserve public legislative scrutiny. Lawmakers are already studying online child safety during the interim, through both a House charge on online gaming platforms and new Senate interim charges.

In written testimony to the Senate State Affairs Committee in September, TPR urged lawmakers to first evaluate how existing laws such as the SCOPE Act and the App Store Accountability Act are working, and to reject broad identification mandates and generalized content controls. If Texas intends to establish broader AI access restrictions or product approval requirements, the 90th Legislative Session, which begins in January 2027, is the place to debate their scope, standards, privacy implications, and application across competing providers.

The Texas Liberty Compact calls on lawmakers to protect free speech and digital expression, recognizing that technological change should not become a pretext for expanding government authority over lawful digital activity. A court-supervised approval process for new AI models is a clear example of that risk: it would place an evaluator, rather than Texans themselves, in the position of deciding which lawful tools are available in the state.

What the OpenAI Lawsuit Means for Texans

The immediate takeaway is that Texas is seeking significant restrictions on ChatGPT, but the lawsuit itself does not put those restrictions into effect.

The case will test both OpenAI's compliance with Texas consumer protection laws and the reach of those laws as a means of governing AI products.

Texas can demand truthful marketing and enforce established legal obligations while preserving parental responsibility and individual liberty. Protecting children and respecting parents' authority are not competing goals, and good policy should accomplish both. The challenge is ensuring that any remedy addresses proven misconduct without creating unnecessarily broad government control over Texans' access to technology.

---

---