• Posted January 22, 2026

Spending A Lot Of Time With AI Chatbots? You've A Higher Risk For Depression, Study Finds

Do you find yourself spending hours chatting with AI programs like ChatGPT, Microsoft Copilot, Google Gemini, Claude or DeepSeek?

Odds are you might be suffering from depression.

People who use AI chatbots daily are about 30% more likely to have at least moderate levels of depression, researchers reported Jan. 21 in JAMA Network Open.

“We found that daily AI use was common and significantly associated with depressive and other negative affective symptoms” like anxiety and irritability, concluded the research team led by psychiatrist Dr. Roy Perlis, director of the Center for Quantitative Health at Massachusetts General Hospital in Boston.

Age also appears to play a role, with middle-aged adults having particularly higher odds of depression if they frequently use generative AI, researchers found.

Regular AI users between 45 and 64 years of age had a 54% higher risk of depression, compared to  32% higher odds among those between 25 and 44, results showed.

Those results indicate that “some individuals may be more apt to experience depressive symptoms associated with AI use,” researchers wrote.

For the new study, researchers surveyed nearly 21,000 U.S. adults between April and May 2025, using a standard mental health questionnaire to track symptoms of depression. Participants were also asked about how often they used AI.

About 10% said they use generative AI daily, including more than 5% who said they turn to it multiple times a day.

From the study design, it’s hard to tell whether AI is promoting depression or if depressed people are turning to AI for solace, researchers said.

Dr. Sunny Tang, an assistant professor of psychiatry at Northwell Health’s Feinstein Institutes for Medical Research in Manhasset, New York, agreed that it’s hard to tell which way the association works.

"People who are already experiencing mental health symptoms may be more likely to use generative AI for personal use by seeking help and support for their symptoms, persuading loneliness, or finding validation,” said Tang, who was not involved in the study.

"When thinking about the link between AI and mental health, we need to think in multiple directions – could AI use negatively impact mental health? But also, how do differences and mental health change the way that we interact with AI?” said Tang, who practices at Zucker Hillside Hospital in Queens, New York.

Loneliness could be an important factor, she said.

“A lot of people are feeling more and more isolated these days, whether it's because they're working remotely or for other reasons,” Tang said. “We know that loneliness is a really strong predictor of mental health symptoms like depression, anxiety and irritability. I think that it's definitely one of the directions where we should be looking to try to understand these relationships.”

The results also showed that AI companies need to produce products that take people’s mental health into consideration, Tang said.

“They should keep in … the forefront the fact that people with mental illness and people with mental health symptoms are going to be actively engaging with their products,” she said. “As all physicians know, first, do no harm." 

Tang said "better guardrails" need to be in place to make sure AI are not providing advice that makes existing mental health symptoms worse.

“Companies should ask themselves, 'Is there a way to build AI so that I can be more supportive of people with mental health needs?’ " Tang said.

More information

The National Alliance on Mental Illness has more on AI and mental health.

SOURCES: JAMA Network Open, Jan. 21, 2026; Dr. Sunny Tang, assistant professor of psychiatry, Northwell Health Feinstein Institutes for Medical Research and Zucker Hillside Hospital