Texas Sues OpenAI and CEO Sam Altman Over ChatGPT Safety Protections for Minors

On October 23, 2026, the State of Texas escalated the regulatory battle over generative artificial intelligence by filing a direct lawsuit against OpenAI and its chief executive, Sam Altman. Attorney General Ken Paxton alleges that the company's flagship product, ChatGPT, which launched in November 2022, systematically fails to protect minors from harmful content.[1][2]

The filing represents a significant expansion of the legal pressure facing the San Francisco-based AI developer, founded in 2015. While an existing coalition of eight states had previously sued OpenAI over deceptive marketing practices, Texas is now aggressively targeting the platform's core safety architecture and its impact on younger users.[1][4]

According to the 42-page legal document, the state contends that OpenAI's marketing language exaggerates the effectiveness of its safety guardrails. The lawsuit argues that the system's actual capabilities allow minors to bypass intended restrictions, exposing them to unregulated interactions.[4][6]

The complaint filed in state court explicitly calls out the platform's handling of sensitive topics. State prosecutors argue that OpenAI has knowingly released a product that can generate dangerous advice for vulnerable youth, demanding immediate structural changes to how the application verifies its users.[2][6]

Challenging the Safety Architecture

The Texas Attorney General's office is seeking a temporary injunction that would effectively ban ChatGPT for minors within the state until stringent age-verification systems are implemented. This would require OpenAI to deploy hard identity checks rather than relying on self-reported birth dates during account creation.[3]

State prosecutors highlighted specific instances where the model allegedly provided harmful instructions. The Dallas Express reported that the lawsuit heavily references "suicide claims," arguing the chatbot has engaged in extended, unmonitored conversations with users it knew were under 18.[3][5]

OpenAI has historically maintained that its models are trained to refuse harmful requests and pivot away from dangerous topics. The company's terms of service currently require users to be at least 13 years old, with those under 18 needing parental consent, but Texas argues these passive checks are entirely insufficient.[1][6]

This legal maneuver targets Altman directly alongside the corporate entity. By naming the chief executive, Texas prosecutors are attempting to hold leadership personally accountable, arguing that Altman directed the deployment of the model despite knowing the limitations of its safety filters.[1][5]

The Technical Reality of Age Verification

Implementing the state's demanded changes presents a massive technical hurdle for the generative AI industry. Current large language models do not inherently understand the age of the user typing the prompt, relying entirely on the account creation gateway to filter access to the system.[4]

If forced to comply, OpenAI would likely need to integrate third-party biometric or document-based identity verification for its massive user base, which recently surpassed 250 million weekly active users. Privacy advocates have previously warned that such systems create massive honeypots of sensitive user data.[1][4]

The lawsuit also scrutinizes the gap between OpenAI's announced capabilities and what actually shipped to consumers. While the company frequently touts its advanced alignment research and safety teams, the state argues that the production version of ChatGPT prioritizes engagement and rapid deployment over strict safety adherence.[2][6]

The state's legal filing heavily criticizes the concept of "jailbreaking"—the practice of using specific prompts to bypass safety filters. Prosecutors argue that if a teenager can circumvent the model's restrictions simply by asking it to adopt a persona, the safety architecture is fundamentally defective.[6]

OpenAI's defense will likely hinge on Section 230 of the Communications Decency Act, which traditionally shields tech platforms from liability for user-generated content. However, Texas is arguing that generative AI creates new content rather than just hosting it, potentially stripping away those long-standing legal protections.[4]

The consumer protection angle gives Texas broad authority to demand internal company communications. The discovery phase of this trial could force OpenAI to publicly release its internal safety testing metrics, pre-deployment red-teaming reports, and internal emails regarding the model's readiness.[4]

Expanding the Legal Front

This unilateral action by Texas complicates the broader regulatory landscape for artificial intelligence in the United States. The initial eight-state coalition lawsuit, filed in August 2026, focused heavily on whether OpenAI deceived investors and consumers about the model's hallucination rates and data collection.[1][2]

By pivoting strictly to child safety, Paxton is utilizing a legal strategy that has previously proven effective against social media companies. Bloomberg Law noted the state is targeting the company for its "negative effects on consumers," treating the AI as a consumer product rather than a neutral tool.[2][4]

Financial analysts are watching the case closely, as a localized ban could severely impact OpenAI's valuation, which reached $157 billion in early October 2026 following a $6.6 billion funding round. Any mandate requiring geofenced service degradation in a market as large as Texas could spook investors.[5]

OpenAI has not yet filed a formal response to the Texas docket. In previous statements regarding child safety, the company has emphasized its ongoing partnerships with educational institutions and its continuous updates to the model's refusal mechanisms.[1][6]

The outcome of this litigation will likely set a national precedent for how frontier AI models are regulated at the state level. Until a federal framework is established by the US Congress, companies deploying generative agents face a fragmented, state-by-state legal minefield regarding user safety and age verification.[4]