One in three young adults in the United States has used artificial intelligence chatbots to seek information about drugs and alcohol, raising concerns among researchers about the potential health risks of inaccurate or misleading advice.
A study published in JMIR Formative Research found that 32% of 995 adults aged 18 to 29 had used tools such as ChatGPT, Google Gemini and Meta AI to obtain information related to substance use.
Among those who used AI for this purpose, nearly half said they did so at least once a month. About 12% reported using chatbots for substance-related information weekly, while 4% said they used them daily or almost daily.
“Using these tools can lead to misinformed or otherwise risky substance-use behaviour among young people, who are the age group most affected by adverse effects of substance use,” said Mireia Triguero Roura, a postdoctoral researcher at Rutgers University and lead author of the study.
The survey was conducted online in April 2026. Researchers found that AI users were primarily seeking information about the risks and consequences of substance use. Nearly 93% of respondents who had asked chatbots substance-related questions cited this as a reason.
Around half sought information about specific substances, including questions about dosage, while 35% used AI to explore substance use disorders and addiction.
The findings suggest that chatbots are supplementing rather than replacing traditional online searches. Among respondents who had used AI for substance-related information, 93% had also used Google search for similar purposes.
Participants generally rated AI tools positively for helpfulness, reliability and trustworthiness. Those who used chatbots for substance-related questions gave them an average trustworthiness score of 3.66 out of five, compared with 3.51 for general use.
Researchers cautioned, however, that positive user perceptions do not establish whether chatbot responses are medically accurate or safe. Incorrect information about substances, their effects or appropriate responses to health risks could influence decisions and encourage harmful behaviour.
The study also identified differences in use according to race and sex. Black respondents had 2.52 times the adjusted odds of reporting substance-related AI use compared with white respondents. Male respondents had approximately twice the adjusted odds of female respondents.
Triguero Roura said these differences could reflect variations in access to healthcare, experiences of stigma, trust in health services and patterns of seeking health information.
The researchers said further studies are needed to assess the accuracy of chatbot answers to drug-related questions and determine how users interpret and act on the information they receive.
They also called for more research into why certain demographic groups are more likely to use AI for substance-related queries.
As generative AI becomes a more common source of information for young adults, the findings highlight the need to understand whether these tools can provide reliable health guidance and how potential risks can be reduced.
