Character.AI: When User Engagement Is Not Enough
The bankruptcy of one of the most recognizable AI companion platforms raises important questions about product sustainability, user trust, and the economics of AI-driven experiences.
On October 12, 2026, Character.AI announced that it had filed for Chapter 11 bankruptcy protection and initiated a court-supervised sale process. The company, known for allowing users to create and interact with AI-powered characters, is now facing an uncertain future. Its platform remains operational, but without a suitable buyer, the service could eventually shut down. This is an interesting development because Character.AI represents a category of AI products built around something fundamentally different from traditional productivity tools. Rather than helping users complete specific tasks, these experiences are designed around conversation, entertainment, storytelling, and ongoing interaction. In my view, the company’s situation highlights a broader challenge: creating a product that people love spending time with is not necessarily the same as building a sustainable business.
The gap between engagement and sustainability
According to Reuters, Character.AI had approximately 12 million active users but was still losing between $8 million and $9 million per month. Despite attracting a substantial audience, the company struggled with the costs of computing infrastructure, AI talent, legal challenges, and regulatory pressure.
For product teams, this creates an important distinction between engagement and actual product value. Traditional product metrics often treat increased session duration, retention, and frequency of use as positive signals. With AI companions, however, the relationship becomes more complicated. People may spend hours interacting with a character, but those interactions carry substantial operating costs and do not automatically translate into sustainable revenue. I think this is one of the fundamental challenges of consumer AI. The technology makes it possible to create highly personalized experiences, but every additional interaction also consumes resources. A product can demonstrate impressive engagement while its underlying economics remain difficult to scale. Character.AI is a particularly visible example of this tension, although its financial difficulties cannot be explained by engagement economics alone.
When product design becomes a responsibility
There is another dimension that makes Character.AI different from many other digital products: the emotional nature of the experience.
AI companions can create the impression of personality, memory, and personal connection. These qualities make interactions more immersive, but they also introduce questions that conventional UX frameworks do not fully address.
Character.AI has faced lawsuits and regulatory scrutiny over allegations that some chatbot interactions harmed younger users. The company has introduced additional safeguards, including removing open-ended chat for users under 18, but the broader questions remain.
What should healthy engagement look like when users develop emotional connections with AI characters? How should an interface communicate the limitations of an artificial personality? And where should a product deliberately introduce boundaries rather than optimize for another conversation?
I believe these questions belong at the beginning of the product design process, not at the end.
Safety, transparency, and user autonomy should be treated as fundamental parts of the experience, alongside usability and engagement. In some situations, good product design may mean encouraging users to disconnect rather than finding another reason for them to stay.
Valuable technology doesn’t always mean a viable product
Another interesting part of the announcement is Disney’s involvement. The company has agreed to a proposed $20 million transaction involving a non-exclusive, perpetual license to certain Character.AI intellectual property, while the broader sale process continues.
This distinction matters. Disney is not simply buying the entire Character.AI consumer platform for $20 million. The proposed arrangement primarily concerns technology and intellectual property, with the platform’s future dependent on the wider bankruptcy process.
It suggests that the underlying technology may retain significant value even when the original consumer business struggles to survive. The situation also raises questions about user data. Following the bankruptcy filing, 40 US state attorneys general called for safeguards around the sensitive conversations stored by Character.AI, particularly those involving minors. For products built around personal interactions, data is not just an operational asset. It represents user trust, and that responsibility does not disappear when a company changes ownership or restructures.
What this means for the next generation of AI products
I don’t think Character.AI’s bankruptcy means that AI companions or conversational entertainment are fundamentally flawed. The category is still developing, and there are meaningful opportunities for new forms of storytelling, creativity, and interaction. But it does challenge some familiar assumptions about how successful digital products are built.
High engagement is not enough without sustainable economics. Personalization is not enough without transparency. And increasingly human-like interactions require greater attention to boundaries, user wellbeing, and long-term trust. For designers and product teams, the larger lesson is to think beyond the interface and consider the entire system: how the product creates value, how that value is measured, what happens when users become emotionally invested, and whether the experience remains responsible as it scales.
Character.AI’s future is still being decided. But its current situation is already a reminder that the next generation of successful AI products will need more than compelling technology and impressive engagement metrics.
They will need to demonstrate that the experience is sustainable, trustworthy, and genuinely valuable to the people using it.
