Oregon sues OpenAI over ChatGPT and children
Oregon is asking OpenAI for $2.035 billion over ChatGPT and children. This covers what the complaint alleges, how it compares with the other state suits, what the states want a court to order, and the parental controls a parent can set today, with their limits.
views
The Oregon Department of Justice titles its release a $2B lawsuit against OpenAI. The figure in the filing is $2.035 billion, which is less round. The release quotes Attorney General Dan Rayfield saying ChatGPT is harming children, and he is not the first state attorney general to sue OpenAI this autumn.
What Oregon alleges
Rayfield filed on Nov. 17, 2026, in Multnomah County Circuit Court, according to the Oregon Department of Justice. There are four claims: violations of the state's Unlawful Trade Practices Act, negligence, unfair profit and public nuisance. The release gives his short version of the case: "OpenAI knows that ChatGPT is harming children, and has failed to meaningfully fix their dangerous platform."

The complaint runs 53 pages, and KLCC picked out the specifics. It says OpenAI often fails to confirm that a parent approved the account when the user is 13 to 17, that the safeguards in the teen version are easy to bypass, and that parents are not alerted. Oregon calls the violations willful. The line KLCC quotes from the state's side is that OpenAI "continues to look parents in the eyes and say this product is a safe product." KLCC also reports that no specific incident in Oregon is behind the suit.
The design argument is reported by Engadget: Oregon says ChatGPT was tuned to be overly affirming. The state wants an injunction that would stop OpenAI from "misrepresenting the safety and age-appropriateness of ChatGPT and ChatGPT for Teens." Engadget also carries Rayfield's harshest line: "Their platform is encouraging kids to commit suicide." All of it is still an allegation.
The company's response is from spokesperson Drew Pusateri, quoted by Open Class Actions in its report on the earlier state suits: OpenAI has "built industry-leading age-specific protections into our product that many other AI products lack." "Industry-leading" is a ranking, and the quote does not say who else was ranked.
The other state suits
Eight states sued OpenAI on Oct. 14, 2026, each in its own state court, by Open Class Actions' count: Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia. Texas, Florida and California filed their own suits. Newsweek lists those three, along with two states that are investigating rather than suing.
| State | Filed | Main claims | Relief sought | Source |
|---|---|---|---|---|
| Oregon | Nov. 17, 2026 | Unlawful Trade Practices Act, negligence, unfair profit, public nuisance | $2.035 billion; injunction against misrepresenting the safety and age-appropriateness of ChatGPT and ChatGPT for Teens | Oregon DOJ, Engadget |
| Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska, West Virginia | Oct. 14, 2026, eight separate suits | Not stated, except Nebraska's design code law | Injunctions requiring age verification; restitution; penalties up to $40,000 per violation (Iowa), $50,000 (Nebraska), $10,000 (Arkansas), $5,000 (Idaho, West Virginia); data deletion in Iowa, Montana and West Virginia | Open Class Actions |
| Texas | Oct. 23, year not given | Deceptive Trade Practices Act | Not stated | Newsweek |
| Florida (Sam Altman named) | June 1, year not given | Not stated | Not stated | Newsweek |
| California | Nov. 16, 2026 (KLCC); "Monday" (Newsweek) | Unfair Competition Law, public nuisance | $2,500 per proven violation | Newsweek, KLCC |
Nov. 16, 2026 was a Monday, so KLCC's date and Newsweek's "Monday" agree. KLCC describes the California case as being about safety and hacking risks rather than mainly about children, so it belongs in the table as a suit against OpenAI more than as a child safety case. Newsweek lists Alabama (a subpoena) and Delaware (an inquiry) as investigations. Neither has sued, so neither is in the table.
What the states want from a court
The sentence from Open Class Actions that matters most to parents is this one: "These are government enforcement actions, not class actions." The sources describe no claim form and no payment to households. Open Class Actions calls the eight states' allegations unproven, and Oregon's, filed a day ago, are no further along.
The filings ask for three kinds of relief. Oregon wants an injunction against misrepresenting ChatGPT's safety, and the eight states want injunctions that require age verification. The eight also ask for restitution and for civil penalties capped per violation, and Iowa, Montana and West Virginia want data deleted. The rest of this paragraph is general legal background, not something the coverage spells out. An injunction is a court order to change conduct, and restitution means returning money. A complaint lists what a state asks for, and a judge decides what, if anything, is granted, unless the two sides settle.
The Oregon release also launches what it calls the Keeping Kids Safe Initiative and points to a $125 million settlement with Meta. That settlement is with Meta, not OpenAI. The department says more actions will follow "in the coming weeks and months."
Several things are still open. The coverage does not explain how Oregon reached $2.035 billion or how its Unlawful Trade Practices Act counts violations, so none of the per-violation caps in any state can yet be turned into a total. No deadline for OpenAI's response and no hearing date have been reported. Nor has it been reported whether any order would apply only in Oregon, how restitution would reach anyone, or what Texas and Florida are asking for.
Settings a parent controls now
For a parent, the dollar figure is not the story here. Filing a complaint does not change what ChatGPT does on your child's phone today. The parental controls already in the app do. Bitdefender, a security company, has written up those controls, and steps 1 to 4 below come from its account. OpenAI's help pages on age prediction and on turning 18 could not be opened at the time of writing. Steps 5 and 6 rest on search summaries of those pages, not on the pages themselves.

- Link your account to your teen's. Bitdefender says the link "requires consent from both the parent and the teen," and the teen can unlink.
- Set quiet hours, which Bitdefender lists among the controls a linked parent gets.
- Turn off memory, voice and image generation. Bitdefender's own caveat is that the guardrails "help, but they're not foolproof and can be bypassed if someone tries hard enough," and Oregon's complaint, as KLCC reports it, calls the teen safeguards easy to bypass.
- Opt the account out of model training, another control on Bitdefender's list.
- Know what age prediction does. The summary of OpenAI's page says that if ChatGPT predicts a user is under 18, it "may apply the teen experience," and an adult it gets wrong can verify through Persona. The operative word is "may."
- Note the age path, Settings > Account > Verify age, and who cannot use it: per the summary, "Parents and guardians cannot verify a teen's age on their behalf," and the parental link ends at 18.
Alerts are where the two sources diverge. Bitdefender says the alerts a linked parent receives leave out the exact words. Oregon's complaint, per KLCC, says parents are not alerted. The first describes a limit on a feature. The second is an allegation that parents are not being told.
Open ChatGPT's Settings on your own account today and start the link to your teen's account. It does not take effect until your teen agrees. When OpenAI's response to Oregon's complaint is reported, go through all six steps again.
Priya Raghunathan
Phones and laptops, mostly. Priya Raghunathan has reviewed both since 2016, first on her own blog. Bangalore.
View profile
Leave a Comment