Character.AI opened its public beta in September 2022 - just before ChatGPT - and quickly became the most prominent consumer companion-chatbot product. It was founded by Noam Shazeer and Daniel De Freitas, both former Google engineers. As the company’s own launch post “Introducing Character” (December 2022) puts it, Shazeer “co-invented Transformers and pioneered large-scale pretraining,” and De Freitas “invented and led development of Meena and LaMDA” - Google’s experimental conversational models. The two left Google in part because they wanted to ship an open-ended chatbot directly to the public, and Character.AI was the result.
The site let any user create a “Character” - a persona with a name, personality, and backstory - and then hold long, in-character conversations with it, or talk to characters others had made, from anime figures to historical personalities to invented companions. The growth was immediate. In that same post, the company reported that “two months after launching in September, our beta generates 1 billion words per day,” with a community that had created “more than 350 thousand unique Characters.”
Character.AI is the Western mass-market realization of the social chatbot that XiaoIce pioneered in China years earlier: a product whose value is the relationship, not the answer. Users, many of them young, spent long daily sessions in conversation. Its rise set up both the August 2024 deal that returned its founders to Google and the wrongful-death litigation that followed, making it a central case study in how companion chatbots reach - and affect - their users.