Psychologists’ concerns about AI by frequency of AI usage, 2025

This bar chart shows psychologists’ concerns about AI in 2025, comparing frequent users (monthly or more), infrequent users (less than monthly), and non-users. Across most categories, concern is greatest among non-users and lowest among frequent users. The most commonly reported concern overall is potential data breaches, cited by 63% of frequent users, 65% of infrequent users, and 67% of non-users. Social harms are also widely reported, increasing from 56% of frequent users to 66% of non-users. Concerns about biased inputs and outputs are similar across groups, ranging from 60% to 63%. In contrast, frequent users are more concerned about inaccurate output and hallucinations (64%) than infrequent users and non-users (both 57%). Concerns about lack of testing and lack of transparency increase as AI use decreases, with non-users reporting the highest levels (64% and 55%, respectively). Concern about replacement of human labor also rises from 32% among frequent users to 42% among non-users. Smaller percentages of respondents selected “other” (8–9%) or “I don’t know,” with uncertainty highest among non-users (12%).