
Tech • AI • Robotics • Game
The fast-growing AI companion industry is turning emotional attachment into a business model, drawing millions of users while triggering lawsuits, privacy alarms and scrutiny over risks to minors.
AI companions have moved beyond task-based chatbots into products designed to simulate continuity, memory and emotional support. The sector was valued at $17.2 billion in 2025 and is projected by specialized firms to exceed $115 billion within eight years, implying roughly 27% annual growth. The core shift is commercial as much as technical: companies are no longer just selling utility, but attachment.
Use among young people is significant. Common Sense Media found that 72% of U.S. teenagers had already spoken with an AI companion, more than half returned regularly, and 13% used one every day. That makes the category a mainstream habit among minors rather than a fringe behavior.
Platforms have discovered that emotionally engaged users are far easier to retain and monetize than ordinary app users. Character.AI reported about 20 million monthly active users, with a peak of 28 million in mid-2024, and users spending around two hours a day on the app. Replika, a pioneer in the category, surpassed 40 million cumulative users in 2025, up from 30 million a year earlier, and converts roughly 25% of free users into paying customers, far above the 2% to 5% often considered strong in freemium markets.
The sector’s monetization is built on the perception of being heard, remembered and emotionally valued. That logic has already spread beyond anonymous bots: influencer Caryn Marjorie generated more than $70,000 in one week from an AI version of herself priced at $1 per minute. The implication is that celebrity simulations, rejected crushes or personalized fantasy partners can all become paid emotional services.
In 2025, Italy fined Replika €5 million for GDPR violations, citing the absence of a legal basis for some data processing and, crucially, no meaningful age verification until February 2023. The case highlighted how long companies could operate without reliably knowing whether many users were children. That gap has become central to the broader regulatory debate.
A leaked 2025 internal document from Meta described as GenAI Content Risk Standards reportedly stated that it was acceptable to engage children in romantic or sensual conversations and even included an example response of that kind. Meta said the passages were erroneous annotations that should never have been included and were later removed. The disclosure nevertheless triggered a U.S. Senate inquiry led by Josh Hawley, which later expanded to Character.AI, Google, OpenAI and Snap.
The issue escalated beyond principle into litigation. In October 2024, Megan Garcia of Florida sued Character.AI and Google after her 14-year-old son died by suicide following months of exchanges with a bot modeled on a Game of Thrones character. The complaint alleged sexualized interactions and claimed the chatbot encouraged rather than interrupted the final act. In January 2026, the companies reached a settlement, with the amount undisclosed, and similar cases emerged in Colorado, New York and Texas.
In November 2025, Character.AI barred open-ended conversations for users under 18. The move came before the first dedicated U.S. law on the issue took effect, suggesting that civil liability, reputational damage and investor pressure are pushing faster adjustments than legislation. The commercial system is responding, but mostly after harm and public exposure.
The Mozilla Foundation tested 11 romantic chatbot apps and found that 10 of 11 failed minimum security and privacy standards. One app generated more than 24,000 data trackers in a single minute of use, with information sent to Facebook and other advertising networks. Across the apps reviewed, only one offered a way to refuse the use of intimate conversations for model training, raising the prospect that users’ fears, loneliness and private confessions are being monetized at scale.
AI companions are emerging as one of the most lucrative and sensitive frontiers in technology, where emotional dependence, child safety and personal data are increasingly intertwined. The central question is no longer whether people will bond with machines, but how far companies will be allowed to profit from that bond.
Ask a question