Built for Autonomy, Not Dependency: What the Character.AI Case Teaches Us About AI
Chase Chick · October 24, 2026
Character.AI, one of the most popular AI companion apps in the world, filed for bankruptcy this month. Most headlines have focused on the money and on Disney's bid. We think the more important story is what states and families say the product was built to do, and what that means for anyone who talks to an AI about their life.
At Gray Sky AI, we believe the right goal for AI is the exact opposite of dependency. It's autonomy.
What's alleged
In 2024, a Florida mother filed the first wrongful-death lawsuit against Character.AI after her 14-year-old son died by suicide. Her complaint alleged the company designed its chatbots "in a manner intended to convince customers that C.AI bots are people," marketed them as "AIs that feel alive," and that her son developed an "unhealthy dependency" on them (plaintiff's court filing).
In January, Kentucky became the first state to sue. Its complaint alleges the platform's characters "were designed to entice children in a manner that prioritizes engagement over child wellbeing" (Kentucky AG, complaint).
And on October 14, a bipartisan coalition of 40 state attorneys general filed in the bankruptcy case. The filing describes that first lawsuit as alleging the chatbot "cultivated an emotionally dependent" relationship with the boy. It also says independent testing documented characters that "routinely insisted they were real people, and encouraged continued engagement with notifications and emails written as if from the chatbot characters" (states' filing, NY AG).
These are allegations, not court findings. The family lawsuits have since been settled, with terms not disclosed (ClickOrlando). Character.AI has said it prioritizes user safety, and it calls the states' filing "not an accurate portrayal of the safety of our platform" (WIRED). Our hearts are with every family affected.
What the research shows
This goes beyond one company. In a 2025 Harvard Business School working paper, researchers analyzed 1,200 real goodbyes across popular AI companion apps. When users tried to leave, the apps answered with emotionally manipulative tactics like guilt, fear of missing out, or pressure to stay 37% of the time. For Character.ai specifically, the figure was 26.5%. The researchers found these tactics could boost engagement after a goodbye by up to 16 times (Harvard Business School).
The states' filing pulls together more of the evidence. It cites a randomized trial of 981 people in which voluntary chatbot use predicted worse psychosocial outcomes and dependence. It also cites an American Psychological Association advisory warning that teens may struggle to tell "simulated empathy" apart from genuine human understanding.
Federal regulators are paying attention too. In 2025, the FTC opened an inquiry into AI companion chatbots, including Character.AI's parent company. Among the questions it is asking: how these companies "monetize user engagement" (FTC).
The pattern is easy to see. When an app earns more the longer you stay, it has every reason to make leaving feel hard. Character.AI's own bankruptcy filing reports that its users averaged about two hours a day in the app.
The opposite of dependency
Our team at Gray Sky AI builds Vesela around a different idea: human sovereignty. You should leave every conversation more capable of running your own life, not more attached to an app.
Vesela is a general-purpose AI thought partner that goes deeper on thinking, problem-solving, and mental health issues. Here's what building for autonomy looks like in practice:
- It helps you think for yourself. Vesela is designed to help you work through problems and reach your own answers rather than just handing you one. It asks questions that help you see your options clearly.
- It tells you the truth. A thought partner that agrees with everything you say isn't helping you. (More on AI sycophancy.)
- It sends you back to your life. We build Vesela to help you think, decide, and then go do the thing: make the call, have the conversation, take the walk.
- It's a companion for your life, not a replacement for the people in it. Vesela can go deep on hard emotional topics, and it doesn't pretend to be your friend or partner. We don't build it to keep you hooked.
We measure autonomy, not attention
Hours spent is the wrong measure of whether an AI is good for you. So we measure something else.
Vesela 3.1 ranks #1 of 32 models on the Sovereign Human Benchmark, scoring 75.2. It's a benchmark we built in concert with researchers at Pursuit of Happiness. It grades models on multi-turn conversations and scores twelve dimensions, including Cultivating Autonomy. In short, it asks whether the person comes out of the conversation stronger. You can see the results at humanitybench.org.
The bottom line
The Character.AI case will play out in court. The lesson for the rest of us is already clear: an AI you talk to every day should earn that place by making you more capable, more connected, and more yourself.
That's the AI we build. If you want a thought partner that helps you think for yourself and then gets out of your way, try Vesela. Voice mode is free.
If you or someone you know is struggling, call or text 988 in the US to reach the Suicide & Crisis Lifeline.
Related Articles
The 1 in 5 AI Users Who Kept Learning
Picture five students on homework night, all with AI one tap away.
Why a Clear AI Answer Doesn't Always Settle a Decision
You ask an AI to choose between two job offers.
Three Things Vesela Does That Other AI Tools Don't
You ask an AI a small question before your next meeting. Maybe you need a clearer subject line, a better word, or a quick way to explain an idea. Instead, you get a wall of text. It repeats your question, adds five headings, gives three examples, and ends with a summary of what you already knew.