wouldit.work

AI tools

Character.AI: a chatbot hit that licensed its models to Google, then filed Chapter 11

Character.AI grew quickly after its 2022 launch, licensed technology to Google as its founders returned there in 2024, and filed for Chapter 11 in October 2026 with Disney as the stalking-horse bidder.

Founder
Noam Shazeer and Daniel De Freitas
Launched
2022
Model
Consumer AI chatbot, with paid subscriptions

Published · 3 min read · 6 sources

Key takeaways

  • Noam Shazeer and Daniel De Freitas launched Character.AI's public beta in September 2022. Two months later they said it was generating 1 billion words per day.
  • In August 2024 the founders returned to Google, and Google signed a non-exclusive license for Character.AI's technology, according to statements given to TechCrunch.
  • On 12 October 2026 Character.AI filed Chapter 11 to run a court-supervised sale. Disney is the stalking-horse bidder, with an auction set for 1 December 2026. If no bid keeps the product going, the company says the platform will be wound down.

What it is

Character.AI is a consumer app where people create and chat with custom AI characters (Introducing Character). It was founded by Noam Shazeer, who has written that he co-invented Transformers, and Daniel De Freitas, who led work on Google's Meena and LaMDA (Introducing Character). The product is still online as of this writing, while the company runs a bankruptcy sale (Character.AI update).

Timeline

  • September 2022: Character.AI launches a public beta. In a 5 December 2022 post, the founders say that two months after launch the beta "generates 1 billion words per day", and that the community had created more than 350,000 characters (Introducing Character).
  • 23 March 2023: Character.AI announces a Series A led by Andreessen Horowitz. It says that in the five months since launch, users had sent over 2 billion messages, with the second billion coming in the previous month, and that active users spent on average over 2 hours a day in the product. It also says creators had already built over 1 million characters (Series A post).
  • 2 August 2024: Shazeer and De Freitas return to Google, with some colleagues, and Google signs a non-exclusive agreement to use Character.AI's technology. Shazeer told TechCrunch he was "confident that the funds from the non-exclusive Google licensing agreement, together with the incredible Character.AI team, positions Character.AI for continued success." General counsel Dominic Perella became interim CEO; the company said most staff were staying (TechCrunch).
  • 18 September 2026: Character.AI says CEO Karandeep Anand is joining Disney as senior executive vice president and chief technology officer, and that members of its technical team are expected to join him. It says the app "is still live and online" (Character.AI).
  • 12 October 2026: Character Technologies, Inc. files a voluntary Chapter 11 case in Delaware to run a court-supervised sale of the company. The board cites "significant legal, regulatory, and operational headwinds." The platform remains up; paid subscriptions stay active for now. If no bid allows continued operations, "the platform will be wound down" (Character.AI update).
  • 4-5 November 2026: the bankruptcy court approves bidding procedures. Disney Worldwide Services is the stalking-horse buyer for certain assets, under an agreement dated 13 October 2026, alongside a transition-services and exclusivity arrangement dated 18 September 2026. Bids are due 24 November 2026, with an auction on 1 December and a sale hearing on 3-4 December (Chapter 11 notice).

Why it worked

  • The product was immediately sticky. Average use of over 2 hours a day, and a billion words a day within two months of launch, showed people would talk to custom characters for a long time (Introducing Character; Series A post).
  • Users made the catalogue. Hundreds of thousands of community-created characters, then more than a million, turned the app into a platform rather than a single bot (Introducing Character; Series A post).
  • The founding team had rare model experience. Shazeer and De Freitas could train and serve their own models at the start, which mattered when few consumer chat products existed (Introducing Character).

What to learn

  1. Early engagement is not the same as a durable business. Character.AI proved demand for character chat, then spent years trying to turn that into a stable company.
  2. A licensing deal and founder return to a big lab can fund the product and also strip out the people who built the models. Compare Inflection / Pi and Adept for similar shapes.
  3. If your product depends on other companies' characters and a young audience, legal and licensing risk is part of the business model, not an afterthought. The company itself pointed to legal and regulatory headwinds when it filed (Character.AI update).

Risks and what didn't work

  • The Google deal did not settle the company's future. Less than two years after the founders left, the CEO joined Disney and the company filed Chapter 11 (TechCrunch; Character.AI; Character.AI update).
  • The sale is not finished. Disney is the stalking-horse bidder as of 5 November 2026, not yet the confirmed owner. The company has said that if no bid supports continued operations, the app will be wound down (Chapter 11 notice; Character.AI update).
  • Headline usage numbers are company-reported. Message counts, word counts and time-spent figures come from Character.AI's own blog posts and are not independently audited (Introducing Character; Series A post).
  • The Series A post does not state the size of the round. TechCrunch later wrote that Character.AI had raised over $150 million, largely from a16z; that is a press figure, not a number from the company's Series A announcement (Series A post; TechCrunch).

FAQ

Short answers drawn from the sections above.

What is Character.AI?

Character.AI is a consumer app where people create and chat with custom AI characters (Introducing Character). It was founded by Noam Shazeer, who has written that he co-invented Transformers, and Daniel De Freitas, who led work on Google's Meena and LaMDA (Introducing Character). The product is still online as of this writing, while the company runs a bankruptcy sale (Character.AI update).

Why did Character.AI work?

  • The product was immediately sticky. Average use of over 2 hours a day, and a billion words a day within two months of launch, showed people would talk to custom characters for a long time (Introducing Character; Series A post).
  • Users made the catalogue. Hundreds of thousands of community-created characters, then more than a million, turned the app into a platform rather than a single bot (Introducing Character; Series A post).
  • The founding team had rare model experience. Shazeer and De Freitas could train and serve their own models at the start, which mattered when few consumer chat products existed (Introducing Character).

What can indie builders learn from Character.AI?

  1. Early engagement is not the same as a durable business. Character.AI proved demand for character chat, then spent years trying to turn that into a stable company.
  2. A licensing deal and founder return to a big lab can fund the product and also strip out the people who built the models. Compare Inflection / Pi and Adept for similar shapes.
  3. If your product depends on other companies' characters and a young audience, legal and licensing risk is part of the business model, not an afterthought. The company itself pointed to legal and regulatory headwinds when it filed (Character.AI update).

What didn't work for Character.AI, and what are the risks?

  • The Google deal did not settle the company's future. Less than two years after the founders left, the CEO joined Disney and the company filed Chapter 11 (TechCrunch; Character.AI; Character.AI update).
  • The sale is not finished. Disney is the stalking-horse bidder as of 5 November 2026, not yet the confirmed owner. The company has said that if no bid supports continued operations, the app will be wound down (Chapter 11 notice; Character.AI update).
  • Headline usage numbers are company-reported. Message counts, word counts and time-spent figures come from Character.AI's own blog posts and are not independently audited (Introducing Character; Series A post).
  • The Series A post does not state the size of the round. TechCrunch later wrote that Character.AI had raised over $150 million, largely from a16z; that is a press figure, not a number from the company's Series A announcement (Series A post; TechCrunch).

Sources

Every link cited above, in order of first use.

  1. 1.Introducing Character (blog.character.ai)
  2. 2.Character.AI update (blog.character.ai)
  3. 3.Series A post (blog.character.ai)
  4. 4.TechCrunch (techcrunch.com)
  5. 5.Character.AI (blog.character.ai)
  6. 6.Chapter 11 notice (blog.character.ai)

Related case studies

More in AI tools →