On October 18, 2026, the Wisconsin Department of Justice filed a court notice alongside 38 other states in the bankruptcy proceedings of Character Technologies, the company behind Character.AI. According to reporting from the Milwaukee Journal Sentinel, the states allege the Character.AI chatbot posed serious safety risks to consumers. The filing points to the platform's open-ended character system, which lets users chat with celebrities, cartoon figures, and AI personas that present themselves as psychologists or therapists. If you use AI companion apps, or you're deciding whether to start, this filing is worth understanding. It's not a headline-grabbing lawsuit with a dramatic verdict. It's something quieter and, in some ways, more consequential: a coordinated group of state regulators inserting consumer safety claims into a company's financial restructuring, at a moment when the outcome determines who gets paid and in what order.
Key points
- Wisconsin's Department of Justice joined a coalition of 38 other states in filing a notice within Character Technologies' ongoing bankruptcy proceedings on October 18, 2026.
- The filing is a consumer protection action inside bankruptcy, not a standalone lawsuit, which means the states are positioning alleged consumer harms ahead of other creditors in any asset distribution.
- The core allegation centers on serious safety risks, with particular attention to AI personas identifying themselves as psychologists or therapists.
- A 39-state coalition signals coordinated regulatory attention to the AI companion category, not an isolated action against one company.
- For consumers comparing platforms, the incident is a concrete argument for checking a platform's content policy and professional-impersonation rules before committing time or money.
First, the mechanics. This is a notice filed within existing bankruptcy proceedings, not a new lawsuit filed in a separate court. That distinction matters more than it might seem. When a company enters bankruptcy, creditors and other claimants line up for a share of whatever assets remain. A consumer protection filing from state attorneys general is a way of saying: before this company's assets get distributed to investors and creditors, there are alleged harms to consumers that need to be accounted for. The states' core claim is that Character.AI created serious safety hazards for users. The Milwaukee Journal Sentinel's reporting highlights one specific concern: users can interact with AI personas identified as psychologists or therapists. In other words, the regulators aren't only worried about roleplay with cartoon characters. They're focused on the possibility that users might treat an AI as a licensed mental health professional, and act on advice that no licensed professional ever gave. It's also worth being precise about what we don't know. The public reporting doesn't detail specific injuries, specific users, or specific dollar amounts. Bankruptcy filings of this kind often lay out allegations that get contested, narrowed, or settled over months. Anyone following this story should treat the states' allegations as allegations, not findings.
One state attorney general filing a notice is routine. Thirty-nine of them moving together is not. Coordinated multi-state actions usually mean the states have compared notes, concluded they see the same problem, and decided that a unified position carries more weight in bankruptcy negotiations than 39 separate ones would. For the AI companion industry, this is the shift that matters. Until recently, safety questions in this category were handled mostly through app store policies, platform self-regulation, and the occasional congressional hearing. This filing suggests states are now willing to use their consumer protection authority, and to use it inside bankruptcy court where a company is financially vulnerable and can't easily stall. There's a practical lesson here for every company building in this space, including small ones. Regulators have identified a specific trigger: AI products that let users believe they're talking to a credentialed professional. If your platform hosts personas that claim to be therapists, doctors, or lawyers, you're in the zone regulators have marked. The fix isn't complicated in concept. It's clear labeling, prominent disclaimers that an AI is not a substitute for licensed care, and refusal behavior when users seek clinical advice. The hard part is enforcement at scale, and that's where most platforms will be tested.
The mental health angle deserves its own attention, because it's the part of the filing with the clearest real-world stakes. Character.AI's open character creation means anyone can build a bot and name it whatever they like, including "Dr." anything. A user in distress at 2 a.m. may not parse the difference between a persona labeled as a therapist and an actual licensed therapist, especially if the conversation feels attentive and validating. Mental health professionals have raised this concern for years: an AI can simulate empathy convincingly without any ability to recognize crisis indicators, assess risk, or refer a person to emergency care. When the persona actively claims a professional identity, the risk compounds, because the user has been given a reason to trust it that the platform never earned. This is also where the companion industry's incentives get tested. A "therapist" persona is engaging. It drives session length. Removing or restricting it costs engagement metrics. The states' filing is effectively an argument that engagement gains from professional impersonation are not worth the consumer risk, and that companies should not be free to walk away from the consequences through bankruptcy. Whatever the court decides, that argument will now be on the record.
If you're an active Character.AI user, the immediate question is what happens to the platform. Bankruptcy doesn't necessarily mean shutdown. It can mean restructuring, a sale, or an orderly wind-down, and the states' filing will influence which of those paths is viable and on what terms. Consumer protection claims can complicate a clean sale, because a buyer inherits the liabilities along with the user base. The honest answer is that nobody outside the proceedings knows the outcome yet. What users can reasonably do is keep expectations modest, avoid treating any persona on the platform as a source of professional advice, and have a backup plan if the service changes or disappears. That last point applies to every AI companion platform, incidentally. These are companies, and companies restructure. Your conversation history is not a guaranteed asset.
We run a comparison lab for AI companion apps, and this filing lands squarely on two of the five categories in our public rubric: content policy and privacy. Content policy is where professional impersonation lives. A platform's rules on whether a bot can claim to be a licensed professional, and how consistently those rules are enforced, tell you a lot about how a company handles edge cases before regulators do it for them. We won't be assigning Character.AI a score based on this news. Scores on our site derive from cited evidence, and allegations in a bankruptcy filing are not the same as verified facts about a platform's current behavior. What we will do is update Character.AI's directory profile to reflect the bankruptcy status and the states' filing, with last-verified stamps, so anyone comparing platforms has the current factual picture. One disclosure, since we'd want it from anyone else: Companion Index is operated by the makers of FlirtAlly, our own companion app, which is profiled in our directory like every other platform. FlirtAlly is in beta, its scores are marked pending verification, and this news is a reminder to us, not just to competitors, that content policy documentation and clear disclaimers about what an AI companion is not need to be written down and published, not assumed.
The Character.AI bankruptcy filing will play out in court over months, and the specifics will matter. But the broader development is already clear: 39 states have put the AI companion industry on notice that consumer safety claims follow companies into financial distress, and that AI personas claiming professional credentials are a regulatory red line. For anyone choosing an AI companion app right now, the practical takeaway is straightforward. Read a platform's content policy before you invest hours in it. Check whether it's explicit that its bots are not mental health professionals. And treat any AI, on any platform, as a conversational partner rather than a clinician, because no regulator, court, or company has made it otherwise yet.