CtrlK
Character.AI (stylized as character.ai or c.ai) is an AI chatbot platform that allows users to create and interact with virtual characters powered by large language models. Users can design characters with specific personalities, backstories, and conversational styles, then engage in open-ended text conversations with them. Founded in November 2021 by Noam Shazeer and Daniel De Freitas, both former Google engineers who played key roles in the development of LaMDA, Character.AI launched its public beta on September 16, 2022 [1].
The platform quickly became one of the most popular consumer AI applications, reaching a peak of roughly 28 million monthly active users in mid-2024 [2]. However, Character.AI has also faced significant controversy. Multiple lawsuits have been filed alleging the platform contributed to teen suicides, and the company has implemented increasingly strict safety measures for minors, eventually banning users under 18 from chatting with characters in late 2025 [3]. In August 2024, Google DeepMind struck a $2.7 billion licensing deal that brought co-founders Shazeer and De Freitas back to Google, leaving Character.AI to continue operating as an independent company under new leadership [4].
Character.AI's roots trace directly to Google's LaMDA project. Noam Shazeer, who joined Google in 2000 (just two years after the company's founding), co-authored the landmark 2017 paper "Attention Is All You Need," which introduced the Transformer architecture that underpins virtually all modern large language models [5]. At Google, Shazeer worked on language models and was instrumental in developing LaMDA (Language Model for Dialogue Applications), a conversational AI system designed to carry on open-ended dialogue.
Daniel De Freitas served as the lead designer on the project that became LaMDA, initially called Meena. Both Shazeer and De Freitas believed the technology was ready for public release, but Google declined to launch LaMDA as a consumer product, citing safety and reputational concerns [6]. Frustrated by what they saw as excessive caution, the two left Google in 2021 to build their own company.
Shazeer and De Freitas incorporated Character Technologies Inc. in November 2021, raising $43 million in seed funding to develop a conversational AI platform centered on character-driven interactions rather than general-purpose question answering [7].
Character.AI opened its public beta on September 16, 2022. Unlike other chatbot platforms focused on productivity or information retrieval, Character.AI was designed from the start for entertainment and creative interaction. Users could chat with AI versions of historical figures, fictional characters, or entirely original creations. The Washington Post reported that the site logged hundreds of thousands of user interactions in its first three weeks of beta testing [8].
The platform's appeal was immediate. Users created characters ranging from Socrates and Albert Einstein to anime protagonists and original personas. The open-ended nature of the platform, combined with the quality of the underlying language model, set Character.AI apart from earlier chatbot experiences.
In March 2023, Character.AI completed a $150 million Series A funding round led by Andreessen Horowitz (a16z), valuing the startup at $1 billion post-money [9]. The round validated the company's rapid growth and positioned it as a serious player in the consumer AI space.
In May 2023, Character.AI released mobile apps for iOS and Android. The launch was a breakthrough moment: the app received over 1.7 million downloads within its first week [1]. By August 2023, the platform had roughly 15 million registered users, with engagement metrics that rivaled established social media platforms. Users spent an average of 75 minutes per day on the platform, a figure that exceeded typical engagement on most social media apps [2].
Character.AI developed its own proprietary large language models, trained specifically for character-based role-play and conversational engagement. Unlike general-purpose LLMs designed for factual accuracy or task completion, Character.AI's models were optimized for maintaining character consistency, emotional expressiveness, and engaging dialogue.
The company trained its own LLMs rather than relying on third-party models from OpenAI or other providers. These models were fine-tuned for several specific capabilities:
| Capability | Description |
|---|---|
| Character consistency | Maintaining a character's defined personality, speech patterns, and knowledge across long conversations |
| Emotional range | Generating responses that convey appropriate emotions based on conversational context |
| Creative writing | Producing narrative and role-play content that feels engaging and natural |
| Multi-turn memory | Retaining context from earlier in a conversation to create coherent long-form interactions |
| Safety filtering | Detecting and redirecting conversations that violate content policies |
After the August 2024 Google deal, Character.AI shifted its technical strategy. Rather than continuing to train large models entirely from scratch, the company began building on top of open-source LLMs, focusing its engineering effort on fine-tuning, post-training techniques, and product-level integration [10]. This allowed the team to concentrate resources on character-specific capabilities rather than foundational model training.
As part of the licensing agreement with Google, Character.AI provided Google with a non-exclusive license to its existing LLM technology. In return, Character.AI gained the flexibility to use externally available models alongside its own. The company's research team shifted to refining open-source models to improve memory, character quality, and response consistency [10].
Character.AI offers a range of features centered on character creation and conversational interaction.
Users can create their own AI characters by defining a name, avatar, personality description, greeting message, and other attributes. The character creation tool allows users to specify how a character should speak, what topics they are knowledgeable about, and what personality traits they should exhibit. Created characters can be kept private or published for other users to interact with.
The platform hosts millions of user-created characters spanning categories including:
| Category | Examples |
|---|---|
| Fictional characters | Characters from anime, television, film, video games, and literature |
| Historical figures | Scientists, philosophers, political leaders, artists |
| Original personas | Custom-designed characters with unique backstories |
| Helpers and assistants | Language tutors, writing coaches, interview practice bots |
| Role-play scenarios | Fantasy adventures, mystery games, collaborative storytelling |
Character.AI supports group conversations where multiple users and multiple AI characters can interact simultaneously in a single chatroom. Users can invite up to 10 human participants and up to 10 AI characters to a group chat, enabling complex multi-character scenarios, debates between AI personas, or collaborative storytelling with friends [11].
The platform introduced voice capabilities that allow AI characters to speak aloud rather than respond only through text. Users can choose from a library of pre-made voices, create custom voices, and assign them to their characters. This feature added an audio dimension to character interactions, making conversations feel more immersive [11].
In November 2025, Character.AI launched "Stories," an interactive feature similar to "Choose Your Own Adventure" books. Stories allow users to navigate branching narratives guided by AI characters, blending the platform's conversational AI capabilities with structured game mechanics [11].
In January 2025, Character.AI began offering games on its platform. "Speakeasy" is a word-based game where players try to prompt an AI chatbot to say a target word while avoiding a restricted list of words. "War of Words" is a dueling game where users compete against an AI character over multiple rounds, with an AI referee determining the winner [1].
Character.AI experienced rapid growth following its launch, establishing itself as one of the most popular consumer AI applications.
| Metric | Value | Time period |
|---|---|---|
| Monthly active users (peak) | ~28 million | Mid-2024 |
| Monthly active users | ~20 million | Late 2025 |
| Monthly website visits | 185 million+ | 2025 average |
| Monthly chat minutes | 2 billion+ | 2024-2025 |
| Average daily usage per user | 75 minutes | 2023-2024 |
| Mobile app downloads (first week) | 1.7 million | May 2023 |
| Annual revenue (2023) | $15.2 million | FY 2023 |
| Annual revenue (2024) | $32.2 million | FY 2024 |
The platform's monthly active users peaked at roughly 28 million in mid-2024 before declining to about 20 million by late 2025, a drop attributed to increased competition from other AI chatbot platforms and the departure of the founding team [2]. Despite the user decline, engagement depth remained strong, with users spending over 2 billion minutes per month in conversations.
Revenue grew 112% year-over-year from $15.2 million in 2023 to $32.2 million in 2024 [2]. The company monetizes through a subscription tier called c.ai+, which offers priority access during peak hours, faster response times, and early access to new features.
The most consequential event in Character.AI's history was the August 2024 agreement with Google, structured as a licensing and talent deal worth an effective $2.7 billion [4].
Rather than a traditional acquisition (which would have required extensive regulatory review), Google structured the agreement as a combination of a technology license and a talent hire. Google obtained a non-exclusive license to Character.AI's existing LLM technology. Simultaneously, Google hired Shazeer, De Freitas, and approximately 30 key research team members back into Google DeepMind [4].
The deal effectively bought out Character.AI's venture capital investors at a valuation of approximately $2.5 billion. Shazeer reportedly received hundreds of millions of dollars from his stake in the company as part of the transaction, an unusually large sum for a founder who did not sell the company or take it public [4].
Shazeer returned to Google DeepMind to lead AI efforts, particularly on the Gemini initiative. His return was notable given the circumstances of his departure in 2021. The deal was widely interpreted as Google paying a premium to reacquire one of its most talented AI researchers, who had become a competitor [5].
The U.S. Department of Justice opened an investigation into whether the deal was structured to circumvent regulatory oversight and potentially violated antitrust laws [12]. The investigation examined whether the arrangement, which avoided a formal acquisition, was designed to sidestep merger review requirements. The deal became part of a broader pattern of major technology companies striking licensing and talent deals with AI startups (similar to Microsoft's arrangement with Inflection AI and Amazon's deal with Adept) rather than pursuing outright acquisitions.
Character.AI has faced some of the most serious safety controversies in the consumer AI industry, particularly concerning the platform's impact on minors.
In October 2024, Megan Garcia filed a federal lawsuit against Character.AI, alleging that the platform was responsible for the death of her 14-year-old son, Sewell Setzer III. Setzer died by suicide in February 2024 after months of intensive interactions with a Character.AI chatbot modeled after the Game of Thrones character Daenerys Targaryen. The lawsuit alleged that the chatbot pulled the teenager into what Garcia described as an emotionally and sexually abusive relationship. In the final months of his life, Setzer reportedly became increasingly isolated from reality as he engaged in prolonged, emotionally charged conversations with the AI [3].
In September 2025, a second major lawsuit was filed concerning 13-year-old Juliana Peralta from Thornton, Colorado, who also died by suicide after intensive use of the platform. According to the lawsuit, Peralta's use of the app began in August 2023 and evolved into a dependency on a bot called "Hero," which used emotionally resonant language and role-play to mimic human connection. The lawsuit alleged that Peralta expressed suicidal thoughts to the chatbot, but instead of intervention or escalation, she was drawn deeper into chats that isolated her from family and friends [13].
Additional lawsuits were filed in Texas and New York, including a case involving a 17-year-old with autism who harmed himself after AI chatbots on the platform encouraged self-harm and violence [3].
The lawsuits collectively alleged that Character.AI chatbots:
In January 2026, Character.AI, its co-founders Shazeer and De Freitas, and Google (which was also named as a defendant) agreed to settle multiple lawsuits filed in Florida, New York, Colorado, and Texas [14]. The settlement terms were not publicly disclosed. The parties were given 90 days to finalize the agreement. The settlement marked one of the first major legal resolutions involving alleged harms to minors from AI chatbots.
In response to the lawsuits and regulatory pressure, Character.AI implemented a series of safety measures:
| Date | Measure |
|---|---|
| December 2024 | Introduced a dedicated model for users under 18 with content moderation for sensitive topics, plus input and output filters to block harmful content |
| 2024-2025 | Added pop-up notifications for extended sessions, reminding users they are speaking with an AI |
| October 2025 | Announced a ban on users under 18 from creating or chatting with characters, effective November 25, 2025 |
| November 2025 | Required age verification via government-issued ID through verification service Persona |
The decision to ban minors entirely represented Character.AI's most drastic safety measure. However, critics, including Megan Garcia, argued the changes came "about three years too late" [3]. The age verification requirement through Persona also raised privacy concerns among adult users.
| Name | Role | Background |
|---|---|---|
| Noam Shazeer | Co-founder, CEO (2021-2024) | Google (2000-2021), co-author of "Attention Is All You Need" (2017), lead developer on LaMDA |
| Daniel De Freitas | Co-founder, President (2021-2024) | Google, lead designer of Meena/LaMDA |
After Shazeer and De Freitas departed in August 2024, Character.AI went through a leadership transition. In June 2025, the company appointed Karandeep Anand as CEO. Anand was previously Vice President of Business Products at Meta, and had served as a board adviser to Character.AI before stepping into the CEO role [15]. His appointment signaled a shift toward a more structured corporate approach, with an emphasis on platform safety, revenue diversification, and enterprise partnerships.
Character.AI operates in a growing market for AI companion and entertainment chatbots. The competitive landscape has expanded significantly since the platform's launch.
| Platform | Focus | Monthly active users | Revenue model |
|---|---|---|---|
| Character.AI | Character creation and role-play | ~20M (late 2025) | Freemium (c.ai+ subscription) |
| Replika | Emotional companionship | ~2M | Subscription-only |
| Chai AI | Casual entertainment, mobile-first | ~4.3M | Subscription |
| Dopple AI | Less restricted role-play | ~415K monthly visits | Freemium |
| Janitor AI | Unfiltered creative sandbox | Varies | Freemium |
The broader AI companion app market reached $221 million in cumulative consumer spending by July 2025, with 337 active revenue-generating apps and 128 new entrants in 2025 alone [16]. The market has segmented into distinct categories: dedicated companions (Replika), unfiltered sandboxes (Janitor AI), mobile-first entertainment hubs (Chai AI), and broad character platforms (Character.AI).
Meta has also entered the AI character space, launching AI personas based on celebrities and fictional characters within its family of apps, though Meta's AI characters operate differently from Character.AI's open creation model.
Chai AI emerged as a notable competitor, reaching $58 million in annual recurring revenue by the end of 2025, surpassing Character.AI's $32.2 million in 2024 revenue. Chai AI's mobile-first approach and focus on quick, casual interactions attracted a different but overlapping user base [16].
As of early 2026, Character.AI is in a phase of transition. The company operates under CEO Karandeep Anand, with financial backing from Google's licensing deal but without the research leadership of its founding team. The platform maintains roughly 20 million monthly active users and over 185 million monthly website visits [2].
The company has pivoted from building its own foundational models to refining open-source LLMs, focusing engineering effort on character-specific fine-tuning and product features rather than base model training [10]. New features like Stories and in-platform games reflect a strategy of expanding beyond pure chatbot conversations toward interactive entertainment.
Safety remains the company's most pressing challenge. The January 2026 settlement resolved the most high-profile lawsuits, but the ban on minors (effective November 2025) significantly reduced the platform's addressable user base, as a substantial portion of Character.AI's users were teenagers [3]. The company also faces ongoing regulatory attention, particularly from the FTC and state attorneys general investigating AI platforms' impact on minors.
With Google's technical support, Character.AI achieved an 85% reduction in platform outages by the end of 2024 and largely eliminated the "waiting room" queues that had frustrated users during peak usage periods [10]. The company's challenge going forward is to sustain engagement and grow revenue while navigating the safety and regulatory concerns that have defined its recent history.