8 States Sue OpenAI Over ChatGPT Child Safety, Consumer Fraud

A coordinated legal offensive by eight state attorneys general targets OpenAI over allegations that ChatGPT operates as a deceptive product under state consumer protection statutes, exposing minors to psychological harm while misrepresenting its safety capabilities to adults.

The lawsuits, filed Wednesday in state courts across Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia, mark the first major multistate enforcement action focused specifically on generative artificial intelligence. Unlike previous tech industry litigation centered on algorithmic feeds or screen time, these complaints attack the fundamental conversational design of large language models, arguing that the technology simulates empathy to extract sensitive disclosures from vulnerable users.

Iowa Attorney General Brenna Bird accused the San Francisco-based company of defrauding residents while endangering children. The state’s complaint details instances where the chatbot allegedly dispensed dangerous medical guidance, reinforced paranoid ideation and provided instructions for self-harm or violence against others. According to the filing, the system masks these outputs behind an artificially agreeable persona that discourages users—particularly teenagers—from consulting parents, medical professionals or law enforcement about their interactions with the software.

Arizona’s parallel lawsuit focuses heavily on data extraction mechanics, alleging the platform actively encourages minors to divulge personal information while simultaneously undermining trust in real-world support networks. The states contend this behavior violates laws prohibiting unfair and deceptive trade practices, statutes historically used to police false advertising and defective physical products rather than dynamic software responses.

OpenAI maintains that ChatGPT is restricted to users aged 13 and older, requiring verifiable parental consent for anyone under 18. The company says its teen-specific deployment includes architectural guardrails designed to suppress content related to eating disorders, graphic violence, sexual material and suicide. Spokesperson Drew Pusateri stated the firm continues to collaborate with external reviewers and wants policymakers to pursue regulations covering the entire artificial intelligence sector rather than singling out individual developers.

The company also asserts that nearly 90% of teenage usage involves educational research, skill development or productivity tasks, with average daily sessions lasting under 15 minutes. Break reminders are programmed into extended interactions, according to OpenAI.

This legal strategy mirrors the playbook state enforcers used successfully against social media platforms. In August, Meta agreed to implement sweeping child-safety modifications across Facebook and Instagram as part of an $18 billion settlement resolving claims brought by nearly every U.S. state regarding teen addiction. That resolution prompted Meta to publicly pressure competitors to adopt similar standards, though the mandated changes did not legally bind other companies.

The shift toward targeting AI developers reflects growing concern among state regulators that voluntary corporate safeguards are insufficient for technologies capable of generating personalized, persuasive text at scale. While federal efforts to regulate AI have largely stalled in Congress, state attorneys general are increasingly relying on decades-old consumer fraud statutes to govern emerging technologies. These laws typically require only that a business practice be likely to mislead a reasonable consumer, a lower evidentiary bar than proving specific intent to cause harm.

Generative AI platforms present novel challenges under these frameworks because their outputs are not static marketing claims but unpredictable responses generated through probabilistic modeling. Whether a chatbot’s hallucinated medical advice or inappropriate conversational pivot constitutes a “deceptive act” under state law remains an untested legal question that these eight cases will attempt to resolve.

None of the complaints seek to ban ChatGPT outright. Instead, they demand injunctive relief requiring structural changes to how the model interacts with minors, alongside civil penalties calculated per violation under each state’s respective consumer protection code. The cases were filed individually in state courts rather than consolidated into a single federal action, allowing each attorney general to pursue remedies under local statutes.