What was filed
On October 13 AI character chat platform Character.AI filed for Chapter 11 bankruptcy protection. Court documents list both its assets and liabilities between $50 million and $100 million. Assets will be sold under court supervision, with closing expected in late November or early December.
Only part of it is settled: Disney has agreed to buy some assets for $20 million and gets a non-exclusive perpetual license to related technology; former Character.AI employees who joined Disney last month, along with former CEO Karandeep Singh Anand, will provide transition services. No one has bid for the company as a whole. Current CEO Colin Adams says the platform keeps running during the sale and he hopes a proposal emerges to keep it alive, "if such a bid is unavailable, however, the platform will be wound down."
From unicorn to bankruptcy
Character.AI was founded in late 2021 by former Google engineers Noam Shazeer (a co-author of the Transformer paper) and Daniel De Freitas, letting users chat with custom characters. It grew extremely fast: a $150 million raise in September 2022 made it a unicorn, its app passed 1.7 million installs in its first week in May 2023, and according to TechCrunch active users averaged more than two hours a day.
The turning point came in August 2024: Google paid $2.7 billion to license its models, both founders returned to Google, and the company stopped training its own foundation models to focus on growth. It reached 20 million monthly users in October 2024. Since then it has faced lawsuits and investigations over the safety of underage users, settled some cases, added safety features and banned minors; it also added AI video generation, social feeds and interactive microdramas to lift engagement, but that didn't make the business sustainable.
Why the math didn't work
The defining trait of AI companion products is that the more users chat, the higher the inference bill, while willingness to pay doesn't necessarily rise with it. Character.AI consistently ranked near the top for time spent, yet couldn't turn that time into enough revenue; after giving up its own models, it also had no technical moat. For teams still building AI character and companion apps, it's a clear reminder to work out inference cost and monetization per active user before chasing time spent. For current users, the platform is still running, but whether it survives depends on the sale, so it's wise to save any conversations that matter to you.