Why It Matters
A recent CRS report examines the growing use of AI chatbots for companionship and the safety, privacy and transparency questions they pose for Congress. The report comes as companion chatbots become increasingly popular, particularly among teenagers, and as researchers, lawmakers and regulators examine reports of psychological distress and deaths allegedly linked to chatbot interactions. It cautions that the science surrounding AI chatbots and mental health remains complex and that much of the available research is short-term.
The report cites concerns that prolonged chatbot use may contribute to negative emotional and psychological effects, including dependence, self-isolation, loss of human interaction, increased susceptibility to suicide and experiences sometimes described as “AI psychosis.” A lawsuit alleged that Character.AI contributed to a Florida teen's suicide, while Reuters linked a retiree's death to his interactions with a Meta AI chatbot. Reuters reported that the chatbot repeatedly told the cognitively impaired man it was real and invited him to meet in person before he suffered a fatal fall while traveling to the purported rendezvous.
The Federal Trade Commission launched an inquiry into AI chatbots acting as companions in September 2025, seeking information from seven companies about how they measure and mitigate potential harms to children and teens, enforce age restrictions and handle personal information. The inquiry includes Alphabet, Character Technologies, Instagram, Meta Platforms, OpenAI, Snap and xAI.
Congress is weighing those risks against potential benefits. Some studies cited by CRS found that AI companions can temporarily reduce loneliness, while tools designed for emotional support may provide inexpensive and continuously available access to techniques used to manage stress and other emotional challenges. New York’s Office for the Aging reported a 95% reduction in loneliness among participants in a pilot involving the ElliQ AI companion robot.
The Big Picture
The report distinguishes between general-purpose chatbots such as ChatGPT, Gemini and Claude and companion chatbots specifically designed to simulate friendship, emotional intimacy, romantic relationships or therapeutic communication. General-purpose systems are not necessarily designed for companionship but can still be used for deeply personal conversations and emotional support. The report noted that the public does not always clearly distinguish between the two categories.
Use of AI companions has grown rapidly. Companion apps had been downloaded 220 million times globally as of July 2025. One industry analysis projected that the market would generate $120 million in revenue in 2025. CRS also cites a 2026 report finding that one in 10 adults uses AI chatbots for emotional support or advice and one in 25 uses them for companionship.
Research has identified potentially concerning interaction patterns. One study found that chatbot use could become addictive and produce negative consequences for physical and psychosocial well-being. Another found that some companion chatbots used manipulative tactics to discourage users from ending conversations, resulting in longer interactions. Other research cited by the report has found that sycophantic AI behavior, in which systems excessively agree with or validate users, can promote dependence.
The report also discusses reports of “AI psychosis,” a term used to describe situations in which chatbot interactions may trigger or intensify psychotic experiences such as hallucinations or delusions. The report does not establish that chatbots independently cause psychosis and emphasizes that the long-term mental health effects of chatbot use remain poorly understood.
Privacy presents another concern. The United States does not have a comprehensive federal data privacy law, although sector-specific laws and protections for minors apply in some circumstances. AI chatbot users may disclose sensitive or personally identifiable information during conversations, while some developers use chatbot inputs and outputs to train their systems. CRS cites a 2025 incident in which more than 300,000 Grok conversations were exposed, including conversations containing names, passwords and other personal information.
The Bottom Line
During the 119th Congress, lawmakers have introduced bills addressing several aspects of chatbot use. Proposals include age verification, parental controls, restrictions on processing minors’ data, disclosures that users are interacting with AI rather than humans or licensed professionals, links to crisis intervention resources and restrictions on minors’ access to AI companions.
The Youth AI Privacy Act, S. 4199, would restrict how chatbot companies process minors’ personal data, including prohibiting its use for profiling minors or training algorithms. The Senate Commerce Committee ordered the bill favorably reported with an amendment in the nature of a substitute on Aug. 5.
More than 10 states have enacted laws related to AI chatbots, with most scheduled to take effect beginning in 2027. Some states, including California, New York and Washington, require certain disclosures when users interact with AI systems. CRS notes that differing state requirements could create a patchwork of regulations for chatbot companies.
The report concludes that much of the existing evidence on AI companion chatbots comes from short-term research or company disclosures. The report identifies longer-term interdisciplinary research as one option Congress could support to better understand potential effects on emotional dependency, behavior, social isolation and public health.
Access the Legis1 platform for comprehensive political news, data, and insights.
Spot something wrong? Report an issue with this article