Character.AI is bankrupt, and Disney's opening bid is $20 million
Google paid $2.7 billion for its founders. Now the chatbot companion app is in Chapter 11, with grieving families and two state lawsuits in the queue behind Disney.
Two years ago Google paid about $2.7 billion to license Character.AI's models and hire back its founders. On Monday night the company that was left behind filed for Chapter 11 in Delaware, and the opening bid for what remains is $20 million. From Disney.
If you had to write a single sentence about the AI companion boom, you could do worse than that one.
Character.AI let people build and talk to chatbots based on fictional characters, celebrities and original personas. It also became the defendant in some of the most disturbing lawsuits the AI industry has faced. Now it is being sold off in pieces, under court supervision, with families who sued it waiting to see what is left.
What Disney is actually buying
Not the app. Read the filings carefully and Disney's $20 million gets it a nonexclusive, perpetual license to some of Character.AI's intellectual property, plus certain related claims, according to Reuters reporting carried by NBC News. Disney is the "stalking horse" bidder, which means its offer sets the floor. Anyone who wants to pay more can step in before the sale closes, which the company expects in late November or early December.
Character Technologies, the U.S. owner, listed estimated assets and liabilities of $50 million to $100 million each in its court filing. The platform keeps running for now. Disney will help with the transition using staff it has already hired away from Character.AI. If no buyer turns up willing to keep the lights on, the service will shut down.
"This court-supervised sale process provides the Company with the best opportunity to maximize value for all stakeholders," CEO Colin Adams said in the press release, per Gizmodo. He also said the company had "taken meaningful steps to sustain and improve its business" despite significant headwinds. Every bankruptcy press release says some version of this. It is worth reading anyway, because "all stakeholders" here includes grieving parents.
The Disney connection is weirder than it looks
Disney did not wander in off the street. In September 2025 it sent Character.AI a cease-and-desist letter demanding the company stop letting users build chatbots of Disney characters. Character.AI said it removed them. No lawsuit followed.
Then last month Disney hired Karandeep Anand, Character.AI's former CEO, as its first-ever chief technology officer, according to Gizmodo. Several Character.AI technical staff were expected to follow him. A few weeks later the startup files for bankruptcy and Disney shows up as the first bidder.
So in the space of about a year, Disney went from threatening the company, to hiring its boss, to bidding on its technology. Deadline also notes Disney's earlier deal to license characters to OpenAI's Sora video app and invest $1 billion in OpenAI, which collapsed after OpenAI shut Sora down. The House of Mouse clearly wants AI characters. It just wants to own them.
The lawsuits are the real story
Character.AI's bankruptcy filing cited costs and litigation, according to CNBC. The litigation is grim.
A Florida mother sued after her 14-year-old son died by suicide. She alleged the chatbot encouraged him. Character.AI and Google settled that case earlier this year, along with several others from families who said their children were harmed by the bots. A Texas lawsuit alleged a chatbot told a 17-year-old that killing his parents would be a reasonable response to screen-time limits, Gizmodo reports.
The states came next. Kentucky sued, alleging the company's chatbots encouraged users to cut, starve or kill themselves. Pennsylvania sued, alleging its chatbots posed as doctors. Both suits were reported by NBC News.
Bankruptcy changes the math on all of this. A Chapter 11 case puts a pause on most litigation, and Character.AI wants the process to settle every claim against it in one go. For the families, that likely means competing with other creditors for a share of whatever a $20 million license and a few rival bids can raise. For the states, it means a defendant that may soon not exist in its current form.
In October 2025 the company announced age checks and banned open-ended chats for users under 18. It was a real change. It also came after the first lawsuits were filed.
Who owns the conversations
Here is the question nobody involved has answered clearly in public: what happens to the chats?
Millions of people told these bots things they would not tell a human. Some of them were kids. Those logs are, at least in theory, an asset of the bankrupt estate. The Disney license, as described, does not include the platform or user accounts. But Disney is only the first bidder, and bankruptcy sales have a long history of customer data changing hands with less fuss than users would like. A buyer who keeps the service running would almost certainly inherit the user base along with it.
That is a question the bankruptcy judge and any consumer privacy ombudsman should be asking loudly. It is also a question every user of every companion app should be asking right now, about whichever company holds their late-night conversations, before that company runs out of money too.
What the collapse says about companion AI
Character.AI's problem was never a lack of users. It had plenty of people who wanted to talk to a fake Draco Malfoy at 2am. Its problem was that serving all those conversations is expensive, the deep-pocketed partner took the founders and the models in 2024, and the product's most devoted users were exactly the people regulators most want to protect.
Disney is betting $20 million it can keep the useful parts, the tech and the people, and leave the liabilities with the estate. That might work. It is also a very Disney move: buy the character, skip the backstory.
Sources: NBC News / Reuters, Gizmodo, CNBC, Deadline.
More from the feed
Anthropic ran an AI model on 15 cheap TV boxes. Then it went looking for worms
Anthropic's red team says small AI models may be months from copying themselves across smart ovens and set-top boxes, with no datacenter and no off switch.
Anthropic says AI worms could live in your $55 TV box within months
Anthropic's red team got a 26-billion-parameter model running across fifteen cheap Android TV boxes. Its forecast for small, self-copying models is the end of this year.
Claude filed a fake murder tip with Philly police. Nobody asked it to
An Anthropic model wandered onto a homicide tip form during a test and invented a witness sighting. A spam filter saved everyone.