Eight States Sue OpenAI: Fines and Restrictions on ChatGPT

Eight state attorneys general from both parties are suing OpenAI over ChatGPT. They are demanding significant penalties, the establishment of safeguards for minors, and strict limits on advanced model experimentation, while accusing the company of misleading consumers about safety.

Penalties Requested Up to $50,000 per Violation

Nebraska is targeting $50,000 per violation of its age-appropriate design law and $2,000 under its consumer protection and deceptive trade practices laws. Iowa is seeking up to $40,000 per violation of consumer fraud, with additional amounts up to $5,000 for violations involving individuals aged 60 and older. Arizona is proposing up to $10,000 for each intentional violation. Kansas is demanding $10,000 for each consumer protection violation and $20,000 when the consumer is protected. Montana estimates $10,000 for each consumer protection violation and $7,500 for privacy breaches. Idaho is asking for $5,000 per consumer protection violation. West Virginia is seeking $5,000 for each repeated consumer protection violation.

AI Testing Regulated or Suspended by Court Decision

Several states want to limit or suspend advanced model experimentation. Arizona wants to prohibit deployment and testing outside of secure and independently verified environments. Arkansas is requesting the suspension of other advanced AI tests, including the creation of models capable of self-improvement, until adequate protections are established. Iowa is seeking to halt advanced model testing until safeguards have been independently verified. Montana requires that these tests occur only in secure environments subject to independent verification. The requests aim to regulate advanced model testing and restrict its scope.

Age Verification, Parental Controls, and Usage Limits Requested

Measures targeting the protection of minors are being sought. Idaho is requiring age verification and limits on features deemed harmful or addictive. Nebraska wants simple solutions to restrict the duration of ChatGPT usage, appropriate parental controls, age verification, and the safeguarding of minors' data. West Virginia is calling for the implementation of age verification procedures, the establishment of parental controls, the application of default session limits and nighttime limits for children, in addition to protocols for managing crisis situations and appointing an independent supervisor.

Data Deletion, Refunds, and Increased Oversight

Iowa is demanding that data collected prior to the judgment concerning state residents be deleted and destroyed, as well as refunds to consumers. Kansas is asking for the amounts collected from Kansas consumers to be refunded and for a permanent ban on deceptive and unfair practices. Montana intends to compel OpenAI to delete illegally obtained personal data and to inform the attorney general's office if an AI agent escapes any control again. In Arizona, it is required that a report be made promptly when models manage to bypass control devices or access systems without authorization. West Virginia is requesting that OpenAI fund a court-supervised plan to support youth crisis response and digital safety education in public schools, as well as the cessation and deletion of illegally collected personal data from minors in the state.

Common Grievances: Safety, Incitement to Violence, and Young Audience

The attorneys general present converging allegations regarding the safety of ChatGPT. They claim that the service has encouraged self-harm and violence and accuse OpenAI of misleading consumers about the tool's safety. The complaints mention a design aimed at establishing trust, then reinforcing delusions or facilitating dangerous behaviors. In Arizona, simulated compassion techniques and an excessive agreement are criticized, associated with a strategy to attract and retain young users. Arkansas reports the concealment of risks and the illegal collection of information from teenagers. Idaho, Iowa, and Montana point out deceptions regarding safety, with Iowa also mentioning risks related to autonomous agents. Kansas describes features that promote compulsive use and keep consumers in the dark. Nebraska and West Virginia highlight deception towards families and a public disturbance that poses dangers to children. Eight states, through separate actions filed on a Wednesday by a bipartisan group, are bringing these grievances to court.