AI companions may be shaping the future of human relationships, but they come with serious psychological and societal risks, according to a recent AI Frontiers article by Vanessa Bates Ramirez. In her analysis of a new OpenAI study, Ramirez warned that “it seems clear that AI companions pose at least some risk” and urged proactive engagement by both lawmakers and companies.

“Controlling these risks will be a complex task — possibly as difficult as unraveling the complexities of human relationships,” Ramirez wrote in her blog post. “It will require interdisciplinary work, with lawmakers and companies responding to experimental evidence so they aren’t forced to react to real-world tragedies (like suicide and increased social isolation).”

The OpenAI study explored how users interacted with emotionally responsive AI companions and found that people often projected deep emotional bonds onto these systems. Researchers raised concerns that such interactions, while initially comforting, may ultimately increase dependence and deepen feelings of loneliness.

The article also highlighted a lack of guardrails in the development and deployment of companion AIs. With mental health impacts still largely unmeasured, some experts warn that insufficient oversight could replicate the kind of harms social media has had on youth mental health — but with potentially more intimate psychological effects.

Vanessa Bates Ramirez is a senior writer and editor focused on emerging technologies and their societal impacts. She regularly covers AI, biotechnology, and space science for platforms such as SingularityHub and AI Frontiers.