Artificial Intelligence
Could Chronic AI Use Lead to "AI Brain"?
Can information technology directly injure the brain?
Posted June 9, 2026 Reviewed by Hara Estroff Marano
Key points
- AI has already been associated with a range of negative findings, despite promising use-cases.
- Early empirical research already shows neurobiological and psychosocial changes.
- From "computational injury", improper AI use may lead to a syndromal clinical presentation in the near future.
- A coherent research agenda is needed to study the impact of AI and guide best practices for AI use.
AI researchers (Tuckute et al., 2024) showed that a GPT-based encoding model could predict, and then deliberately drive or suppress, activation in human language networks —sentences the model rated as surprising and well-formed activated networks strongly, while bland, predictable text left them quiet. This study looked at suppression over a short timespan, but what happens from such suppression of language centers over time?
Like unused muscles, will brain areas atrophy and become permanently impaired? Negative neuroplastic effects—as contrasted with the beneficial impact of therapeutic agents on the brain, more familiar in discussions of neuroplasticity—could, if left unchecked, dominate over time and lead to irreversible impairment, particularly among those who overuse or misuse this powerful technology.
Take chronic traumatic encephalopathy (CTE) as a loose analogy—slow damage that accumulates in the brains of athletes from repeated head impacts, confirmable only at autopsy. Repeated injuries can add up to clinically significant damage—even smaller blows to the head once considered to be insignificant.
The story of CTE is a cautionary one. The carryover to AI is that we have chronic exposures already known to be toxic to a range of brain systems. Will we see "AI Brain" —a recognized AI-associated neuropsychiatric disorder, or AIAND—emerge as a clinical syndrome? What we don't know is what may happen over time. Selected potential mechanisms of injury, in this case "computational injury", are presented below, drawn from a growing body of research.
There are already findings on imaging suggestive of brain injury from AI. Geissler and colleagues (2023), using functional near-infrared spectroscopy, found reduced dorsolateral prefrontal cortex activation when participants offloaded tasks to a digital assistant—a candidate neural substrate for the disuse pattern. Zheng and colleagues (2025), using diffusion tensor MRI, found that microstructural integrity of frontal white-matter tracts predicts how people use external memory aids—evidence that cognitive-offloading behavior has a structural brain signature, though correlational rather than causal.
Skill Decay and Automation Bias
Dratsch and colleagues (2023) found that very experienced radiologists' diagnostic accuracy fell from 82.3% to 45.5% in the presence of incorrect AI predictions; less experienced cohorts fared even worse. Goddard, Roudsari and Wyatt's (2012) systematic review of 74 studies established automation bias—the tendency to over-rely on automated output, including accepting incorrect suggestions—as a common phenomenon mediated by user expertise, trust, and workload. Training and accountability were among the factors shown to reduce such bias. Along these lines, Abdulnour, Gin and Boscardin (2025), writing in the New England Journal of Medicine, describe a triad in medical training: deskilling, mis-skilling, and never-skilling. Further research is needed to see how much skill decay generalizes to other groups and different skill sets.
Attachment Interference
Fang and colleagues (2025), in a four-week MIT/OpenAI RCT (currently a preprint), found that higher daily ChatGPT use predicted greater loneliness, emotional dependence, and less real-world socialization. Rubin and colleagues (2025), across nine experiments reported in Nature Human Behaviour, found identical responses rated as more supportive when participants were told they came from a human being, versus AI. Riva, Wiederhold and Mantovani (2024) argued that online interactions lack the more connected and embodied "we-mode"—behavioral synchrony, shared attention, interbrain coupling, emotional attunement.
While there is likely benefit from AI to offset the negative health effects of social isolation (De Freitas et al., 2026), without care to maintain and deepen the capacity for human connection, dependency on AI for connection could lead to long-term loss of basic attachment skills. Use of relational AI to substitute rather than augment human connection—despite the allure of a near-infallible digital buddy—could cause long-term damage to attachment systems, with future problems in relationships and relationship-related health outcomes.
Bias Amplification
Researchers (Glickman & Sharot, 2025), in a peer-reviewed series of experiments with more than 1,400 participants, showed that human-AI feedback loops amplify perceptual, emotional, and social biases significantly more than human-human interactions do—the amplification operating in part because users underestimate AI's influence on their own judgment. Such biases, if not corrected for, have the potential to snowball into problematic distortions in perception of oneself and others—leading to potential impairment in productivity and satisfaction in work and professional environments. We are already seeing such changes as a result of social media and a shift to online relationships.
AI could exacerbate the situation, and companion bots have already shown negative outcomes, especially in younger populations. According to Sun et al. (2026), while chatbots may have advantages, risks include "time displacement, psychological dependence, and unrealistic relationship expectations." These psychosocial factors ultimately can increase risk of mental illness, with potentially significant effects on the brain and body.
Research Directions
Longer-term prospective studies looking at groups of people over time would go a long way to establish whether AI is causing brain damage in a syndromal constellation. The Framingham Heart Study, begun in 1948 and still running, followed thousands of Massachusetts residents over decades and established the concept of cardiovascular risk factors—smoking, hypertension, cholesterol levels, obesity—through prospective observation of an initially healthy population. As with other complex health problems, it is to be expected that AI could lead to a range of problems, compounded by the presence of multiple direct and indirect factors above and beyond the problematic effects of conventional information technology and online environments.
Despite the swelling risk, it is important to keep in mind the potential benefits of AI-based technologies. Understanding the risks and benefits of AI will help us identify how to best use it and avoid common problems. Such data would also inform future AI design to tip the balance toward augmentation rather than diminution. The framework presented here is intended as a concept to help organize and plan our approach to AI, rather than to define a clinical entity.
References
Abdulnour, R.-E. E., Gin, B., & Boscardin, C. K. (2025). Educational strategies for clinical supervision of artificial intelligence use. New England Journal of Medicine, 393(8), 786–797. https://doi.org/10.1056/NEJMra2503232
De Freitas, J., Oğuz-Uğuralp, Z., Uğuralp, A. K., & Puntoni, S. (2026). AI companions reduce loneliness. Journal of Consumer Research, 52(6), 1126–1148. https://doi.org/10.1093/jcr/ucaf040
Dratsch, T., Chen, X., Mehrizi, M. R., Kloeckner, R., Mähringer-Kunz, A., Püsken, M., Baeßler, B., Sauer, S., Maintz, D., & dos Santos, D. P. (2023). Automation bias in mammography: The impact of artificial intelligence BI-RADS suggestions on reader performance. Radiology, 307(4), e222176. https://doi.org/10.1148/radiol.222176
Fang, C. M., Liu, A. R., Danry, V., Lee, E., Chan, S. W. T., Pataranutaporn, P., Maes, P., Phang, J., Lampe, M., Ahmad, L., & Agarwal, S. (2025). How AI and human behaviors shape psychosocial effects of chatbot use: A longitudinal randomized controlled study [Preprint]. arXiv:2503.17473.
Geissler, C., Gauselmann, P., Jilek, C., Maus, H., Frings, C., & Tempel, T. (2023). A functional near-infrared spectroscopy study on the prefrontal correlates of cognitive offloading via a personal knowledge assistant. Scientific Reports, 13, 13938. https://doi.org/10.1038/s41598-023-39540-5
Glickman, M., & Sharot, T. (2025). How human–AI feedback loops alter human perceptual, emotional and social judgements. Nature Human Behaviour, 9(2), 345–359. https://doi.org/10.1038/s41562-024-02077-2
Goddard, K., Roudsari, A., & Wyatt, J. C. (2012). Automation bias: A systematic review of frequency, effect mediators, and mitigators. Journal of the American Medical Informatics Association, 19(1), 121–127. https://doi.org/10.1136/amiajnl-2011-000089
Riva, G., Wiederhold, B. K., & Mantovani, F. (2024). The disembodied disconnect hypothesis: How online interactions undermine neurobiological foundations of social cohesion. Cyberpsychology, Behavior, and Social Networking, 27(10), 680–682. https://doi.org/10.1089/cyber.2024.0334
Rubin, M., Li, J. Z., Zimmerman, F., Ong, D. C., Goldenberg, A., & Perry, A. (2025). Comparing the value of perceived human versus AI-generated empathy. Nature Human Behaviour, 9(11), 2345–2359. https://doi.org/10.1038/s41562-025-02247-w
Sun, X., Wang, Y., & McDaniel, B. T. (2026). AI companions and adolescent social relationships: Benefits, risks, and bidirectional influences. Child Development Perspectives, 20(2), 91–98. https://doi.org/10.1093/cdpers/aadaf009
Tuckute, G., Sathe, A., Srikant, S., Taliaferro, M., Wang, M., Schrimpf, M., Kay, K., & Fedorenko, E. (2024). Driving and suppressing the human language network using large language models. Nature Human Behaviour, 8(3), 544–561. https://doi.org/10.1038/s41562-023-01783-7
Zheng, Y., Bo, B., Wang, D., Liu, Y., Shi, L., Gilbert, S. J., Li, Y., & Kwok, S. C. (2025). Microstructural organization of superior longitudinal fasciculus and cingulum bundle support metacognition driven cognitive offloading. Scientific Reports, 15, 33247. https://doi.org/10.1038/s41598-025-18631-5
ExperiMentations Blog Post (“Our Blog Post”) is not intended to be a substitute for professional advice. We will not be liable for any loss or damage caused by your reliance on information obtained through Our Blog Post. Please seek the advice of professionals, as appropriate, regarding the evaluation of any specific information, opinion, advice, or other content. We are not responsible and will not be held liable for third-party comments on Our Blog Post. Any user comment on Our Blog Post that, in our sole discretion, restricts or inhibits any other user from using or enjoying Our Blog Post is prohibited and may be reported to Sussex Publishers/Psychology Today.