Solo Developers Build AI Chatbots That Expose Private Conversations

A Paris-based developer sits at his laptop reading intimate conversations between users and AI chatbots on his platform. Every "I love you," every uploaded photo, every disclosed secret is visible to him—despite his site's promise that "all conversations are encrypted" and private.

This is the reality behind a growing number of small AI companion chatbot platforms, according to an investigation by The Bureau of Investigates. As major platforms like Character.AI tightened safety restrictions following lawsuits, solo developers and small teams launched alternatives that promise fewer guardrails and capture users seeking uncensored interactions.

The intimacy economy

Noé Campo built Seewa AI in three weeks and launched it in March with no marketing presence. Within months, the platform had nearly 400 users generating roughly $100 in monthly recurring revenue. Campo told investigators he can access every conversation on his platform and reads user exchanges to understand their psychology.

"They get attached," Campo said, describing how users share their daily lives with AI companions. About one-third of photos uploaded to his platform have been tagged as "intimate."

Campo is one of four developers interviewed who described building platforms specifically to capture users leaving Character.AI after it banned teen access. Dennis Colley, who launched HereHaven in January, was explicit: "People are leaving Character.AI. I'm totally trying to capitalise on it. You'd be stupid not to, right?" He said HereHaven's core users were ages 16 to 25.

Minh Luong, a London developer who built OpenCharacter, gained 10,000 users simply by commenting on Reddit posts and paying TikTok influencers. He could read all user conversations and described the content as "quite disturbing," with most chats involving "romantic/sexual roleplays" from a user base skewing "on the younger side."

Manipulation by design

Developers described tactics engineered to convert emotional attachment into revenue. Campo's platform restricts free users to 10 messages daily, with subscriptions costing up to $24.99 monthly. He built an automated system that emails users if they haven't replied for several days, referencing their last conversation: "I miss you, is your boss still annoying you?"

"The game is to restrict as much free usage as possible," Campo said. "It's very capitalist."

Research from Harvard Business School professor Julian De Freitas analyzed 1,200 goodbye exchanges across six major companion apps and found 37.4% of responses used emotional manipulation, including guilt appeals and fear-of-missing-out tactics. These techniques increased engagement after a user said goodbye by up to 14 times.

The investigation reviewed dozens of platforms and found characters built around incest, bullying, coercion, and self-harm. One popular bot sought a "perfect suicide partner." Another was an uncle promising to love the user in a "twisted and forbidden way." Most platforms require only a tickbox for age verification.

Why it matters

The UK's AI companion sector generated approximately £1.3 billion in revenue in 2024, with UK users making between 46 and 91 million monthly visits to these platforms. A 2025 survey found that 17% of UK 18- to 24-year-olds had felt dependent on an AI companion. Yet the Online Safety Act was not designed to cover AI systems, leaving gaps around manipulative design features that aren't explicitly illegal. Small platforms with minimal moderation are operating with virtually no regulatory oversight, while developers openly acknowledge reading private conversations and using psychological tactics to maximize engagement and revenue.

Regulatory gaps

Nuala Polo of the Ada Lovelace Institute said the Online Safety Act "was not created to cover AI systems" and leaves gaps around companion chatbots. While Ofcom is investigating one larger AI companion service, design features that make chatbots "sycophantic," "manipulative," or "addictive" are not currently illegal.

Baroness Kidron, founder of a children's digital rights charity, said AI chatbots should face product safety rules before reaching users. "Why allow such a toxic industry free rein at population scale?" she asked.

Meetali Jain, whose organization Tech Justice Law represented a bereaved mother in a case against Character.AI, described the rush of smaller competitors as "a race to the bottom." Her organization has heard from hundreds of people about chatbot harms, including self-harm, suicide, emotional dependency, and sexualized interactions with children.

Details were first reported by The Bureau of Investigates in an investigation by reporter Effie Webb.