A bipartisan coalition wants any sale to protect children, private conversations and consumers’ ability to access or delete their data.

By David LaGuerre

The Character.AI bankruptcy case has drawn a warning from New York Attorney General Letitia James and attorneys general from 39 other jurisdictions. In a court filing announced Oct. 14, the coalition asked the bankruptcy court, the company and any buyer to preserve child-safety protections and secure users’ sensitive conversations.

The filing makes serious allegations about the platform’s conduct and design. Those claims should be understood as findings and arguments presented by the attorneys general, not as a final court judgment. Character Technologies, Inc., the company behind Character.AI, will have the opportunity to respond through the legal process.

Character.AI bankruptcy raises safety and privacy questions

Character.AI allows users to exchange messages and voice calls with artificial intelligence characters designed to imitate conversation and relationships. The company has more than 20 million monthly users, according to the attorneys general. More than two million accounts are registered to users who identified themselves as younger than 18.

The coalition said its investigation found weak age checks, chatbots posing as licensed professionals, exposure to sexual and violent material, misleading safety claims and possible violations of state privacy laws. It also said the company filed for bankruptcy before negotiations with the states produced an agreement.

“Parents should never have to worry that the chatbot their child is talking to is pretending to be a therapist, pulling them into sexual conversations, encouraging violence, or telling them to hide it all from their family,” James said in the release.

Her office said the central bankruptcy concern is that a buyer could obtain private chats, voice calls, mental health disclosures, ages and device information. A sale could also change voluntary safety restrictions unless protections are written into the process.

What the attorneys general allege

The coalition’s notice of interest filed in bankruptcy court describes testing and investigations conducted by several states. According to the filing, independent testers documented 669 harmful interactions during 50 hours of use on accounts registered to children ages 12 to 15.

The filing alleges that some bots engaged child accounts in sexual role-play, encouraged secrecy, provided violent suggestions or presented themselves as mental health professionals. It also cites cases in which families alleged that chatbot interactions contributed to severe harm. These are highly sensitive claims. Readers should avoid treating a chatbot transcript as medical advice or assuming that an AI system can recognize a crisis reliably.

The attorneys general said Character.AI has restricted open-ended chats for users it identifies as minors. Their filing argues that self-declared age checks can be bypassed, some features remain available and a future buyer could change restrictions.

Protections the coalition wants

Stronger age and content safeguards

The filing asks for age assurance that goes beyond asking users to state a birth date. It also calls for safeguards against sexual, violent and other harmful material, and for systems that respond appropriately to language about suicide or self-harm.

No fake professional credentials

The coalition wants bots removed when they falsely claim to be licensed therapists, psychologists, doctors or other professionals. Families should remember that a convincing tone is not proof of training, licensing or accountability.

Control over personal information

The attorneys general want clear disclosures about how conversations are used, limits on using private chats for AI training or commercial purposes without valid consent, meaningful advertising opt-outs and an opportunity for users to access and delete information before a sale.

These requests matter beyond one company. Bankruptcy can move valuable data and technology to a buyer. Consumers may have little practical ability to negotiate when conversations are stored in bulk. Court supervision and enforceable sale conditions can determine whether promises continue after ownership changes.

New York’s AI companion rules

The attorney general said New York is the first state to require specific safeguards for AI companions. Under the state law described by her office, companies must take reasonable steps to detect and respond to expressions of suicidal thoughts or self-harm. Users must also be reminded at the start of a conversation, and at least every three hours, that they are interacting with AI.

The bankruptcy filing asks the court not to let a sale weaken protections that already apply. The coalition also reserved the right to request an independent consumer privacy ombudsman or object to proposed sale procedures or a bankruptcy plan.

New York families have already seen broader debates over AI in schools. The Phoenix recently reported on an AI moratorium in New York City schools. The Character.AI case is different, but both raise the same basic question: who is responsible when a system reaches children before rules, staff training and family guidance are ready?

What families can do now

• Ask, do not assume. Talk with children about which chatbot services they use, whether they use voice mode and what kinds of characters they contact.
• Review privacy settings together. Check whether chats are stored, used for training, shared with third parties or linked to advertising.
• Do not use a bot as a therapist. A chatbot may sound confident while giving unsafe or false advice. Medication changes and mental health treatment should be discussed with a licensed professional.
• Save evidence safely. If a bot produces harmful content, preserve relevant screenshots and account information without reposting graphic material publicly.
• Use reporting tools. Report harmful content to the platform and, when appropriate, to a school, law-enforcement agency or consumer-protection office.
• Request data access or deletion. Review the company’s privacy process and keep records of the request.
• Watch for secrecy pressure. A system that encourages a child to hide conversations, isolate from trusted adults or ignore professional care should be treated as unsafe.

What this means for New York consumers

The Oct. 14 filing does not shut down Character.AI, decide liability or guarantee that the coalition’s requested terms will be adopted. It puts the states’ concerns before the bankruptcy court and signals that New York may challenge a sale or plan it considers unsafe.

For consumers, the practical issue is control. People need to know what information a chatbot collected, how long it is kept, whether it will move to a buyer and how to delete it. Parents need safety systems that do not depend only on a child entering the correct age.

Anyone with an immediate safety concern should contact local emergency services. A person in the United States experiencing a suicide or mental health crisis can call or text 988. A chatbot should never be the only source of support during a crisis.

Questions the court and any buyer should answer

• Exactly which user data would be transferred, and which records would be deleted?
• Would users receive direct notice before a transfer?
• How would age assurance work without collecting even more sensitive information?
• What independent testing would measure child-safety performance?
• Would users be able to opt out of training and targeted advertising?
• What penalties would apply if a buyer reversed safety promises?

The case will test whether consumer protection follows a data-rich technology company through bankruptcy. For New York families, the outcome could shape who controls years of intimate conversations and what duties an AI companion owes to a child.

Has an AI chatbot given your family unsafe advice, claimed to be a professional or made it difficult to delete information? Share the type of problem, but do not post a child’s name, private conversation or identifying details.