40 attorneys general are watching the Character.AI bankruptcy: conversations with minors aren't just another asset

By 3B Media News · October 23, 2026.

A coalition of 40 U.S. attorneys general, led by Tennessee's Jonathan Skrmetti, has filed a formal document in the bankruptcy proceedings (Chapter 11) of Character Technologies, Inc., the creator of the Character.AI chatbot. The filing is a «Notice of Interest and Reservation of Rights».

According to 3B Media News, the goal is to ensure the company cannot evade consumer protection laws or accountability to regulators during its restructuring or a possible sale of assets. Chapter 11 is the U.S. procedure that allows an insolvent company to reorganize or be sold off in parts.

The attorneys general, according to the same source, point to two areas of concern. The first is the platform's safety practices. The second is its handling of sensitive data: they argue that Character.AI has collected highly personal conversations, disclosures about mental health and private information from millions of users, including children. The filing also warns that companion chatbots can foster emotional dependence and expose minors to risks. Skrmetti said AI developers must be held accountable for user safety and privacy regardless of their financial situation.

The available material is sparse. It does not detail figures from the bankruptcy, the court schedule or potential buyers. Nor does it specify which particular practices are being questioned. What follows is therefore our take, not a summary of the case file.

**Our take.** When a company goes bankrupt, its data becomes an asset that can be sold to the highest bidder. Here the asset consists of intimate confessions from vulnerable people, and the question is whether the buyer will inherit the obligations under which they were collected. The attorneys general are trying to make the answer yes: that liability travels with the data and does not stay behind in the bankrupt entity. It is worth remembering that, for now, this is a reservation of rights and not a proven accusation: the concerns are those of the attorneys general, and the material records no judicial findings.

In general, the companion AI sector is funded through subscriptions and microtransactions and competes to retain users. That creates incentives that sit poorly with protecting minors. If the market does not sustain safety, regulation ends up arriving by the least comfortable route: through the courts and the states, as is happening here.

The case ties into a thesis we keep repeating: a safety commitment is only worth as much as its consequences. A bankruptcy is the acid test, because it is the moment when the economic incentive to comply disappears. That 40 states are acting at once suggests a political consensus, rare in the U.S., on protecting minors from this technology.

In the short term, the transition is uncomfortable: we have deployed artificial companions at scale without properly understanding their emotional effects and without clear rules on data. In the long term, AI that accompanies, teaches and cares for people may be one of the great advantages of the abundance we are pursuing. But getting there requires that trust be built before business, with data treated as a trust deposit and not as liquidatable inventory.