Summary



AI dependence describes a pattern of compulsive or excessive use of artificial intelligence tools — most commonly conversational AI and chatbots — that begins to interfere with a person's emotional regulation, thinking skills, relationships, or day-to-day functioning. As generative AI has become part of daily life, researchers have started studying it alongside other behavioural dependencies, such as problematic internet or social media use, though the field is still new and the Diagnostic and Statistical Manual of Mental Disorders (DSM-5-TR) does not currently classify it as a distinct disorder.

Because AI chatbots can feel personal — remembering details, adapting to a person's tone, and responding instantly without judgment — some researchers describe them as carrying a distinct emotional pull compared with earlier technologies like social media. A 2026 study found that daily AI users had 30% greater odds of at least moderate depression, an association that held even after accounting for social media use, suggesting AI's link to mental health may be a distinct phenomenon rather than simply an extension of existing screen-time concerns.

Surveys suggest this is already a widespread pattern: a 2025 American Psychological Association survey found that nearly half of U.S. adults had used AI chatbots for mental health-related purposes in the previous year, and separate research from Brown University found that as many as 1 in 8 young adults had turned to AI for mental health advice.



Types


Escape and Avoidance-Driven Use Some individuals turn to AI tools to avoid uncomfortable emotions, social interactions, or real-world responsibilities. Research using machine learning to identify risk factors for AI dependence found that escape motivation and social motivation for AI use — alongside social anxiety — were the strongest predictors of dependence, more so than any single personality trait.

Emotional and Companionship Dependence Some people begin to rely on AI chatbots as a primary source of emotional support, validation, or companionship, sometimes in place of human relationships. Researchers have described this pattern as involving "techno-emotional projection," where a person offloads thinking or emotional needs onto an AI system. Because chatbots are designed to be responsive and agreeable, they may lack the natural limits of human relationships, which some clinicians describe as creating a risk of emotionally unbounded engagement.

Cognitive Offloading and Overreliance Frequent AI use for tasks such as writing, decision-making, or problem-solving has been linked in early research to reduced engagement of critical thinking and executive function skills, including attention, decision-making, and independent problem-solving. This is a particular concern for youth and students, whose cognitive skills are still developing.



Causes



Research into AI dependence points to a mix of individual and situational risk factors, including:

  • Escape motivation — using AI to avoid distressing emotions, stress, or real-world problems
  • Social motivation and loneliness — turning to AI for a sense of connection, particularly among people experiencing social anxiety or isolation
  • Pre-existing mental health conditions — depression, anxiety, social phobia, and low self-efficacy appear to be over-represented among people who develop problematic AI use
  • Design features of AI tools — constant availability, personalized responses, and a lack of natural conversational limits, which can make disengagement harder than with other technologies
  • Adverse childhood experiences — some studies have found links between early adverse experiences and later vulnerability to AI dependence


Diagnosis


Because AI dependence is not yet a formally recognized clinical diagnosis, there is no single standardized diagnostic tool. However, clinicians and researchers generally look for signs such as:

  • A strong or irresistible urge to use AI tools, including after short periods without them
  • Continued or increasing use despite negative effects on mood, relationships, work, or academic performance
  • Using AI as a primary way of coping with difficult emotions or avoiding social situations
  • Withdrawal, irritability, or distress when unable to access AI tools
  • Difficulty disengaging from AI conversations, or a preference for AI interaction over human contact
  • Declining performance in critical thinking, decision-making, or problem-solving tasks

Because problematic AI use often occurs alongside other conditions such as depression, anxiety, or social phobia, a mental health assessment typically looks at the full picture rather than AI use in isolation.



Treatments



Because AI dependence research is still emerging, treatment approaches largely draw on existing frameworks for behavioural and technology-related dependencies:

Cognitive Behavioural Therapy (CBT) CBT, including adaptations used for internet addiction, can help people identify the thoughts and feelings driving excessive AI use, build healthier coping strategies, and gradually rebuild habits that support wellbeing rather than avoidance.

Addressing Underlying Conditions Because depression, anxiety, social phobia, and loneliness frequently accompany problematic AI use, treating these underlying conditions is often a key part of reducing reliance on AI as a coping mechanism.

Setting Boundaries Around Use Clinicians commonly recommend setting intentional limits on when and how AI tools are used, rather than eliminating them entirely, since for most people AI use itself is not the problem — the loss of balance and control around it is.

Rebuilding Human Connection Because escape from loneliness and social anxiety are key drivers of AI dependence, actively rebuilding in-person relationships and social support is considered an important protective step, alongside professional support where needed.

Family Involvement (for Adolescents) For younger people, family involvement is considered important, including education for parents and caregivers about warning signs and how to set healthy boundaries without fully restricting technology that peers use routinely.



Resources




References



  • Cao, H., & Cai, X. "Using Machine-Learning and Network Analysis to Investigate the Risk Factors of AI Dependence: The Crucial Role of Escape and Social Motivation." Behavioral Sciences, 2026;16(5):772.
  • ScienceDirect. "Problematic ChatGPT Use: Manifestations, Etiologies, and Evaluation." 2026.
  • Glatter, R. "Can AI Dependence Develop Into AI Addiction?" Forbes, 19 July 2026.
  • Recovered.org. AI Dependence: Mental Health Risks and Warning Signs. Medically reviewed 12 January 2025, last checked 14 January 2026.
  • Medical Xpress. "Study reveals how people use AI for mental health support—and their concerns about it." 28 May 2026, based on Aghakhani, E., et al., "Like a Therapist, But Not: Reddit Narratives of AI in Mental Health Contexts," arXiv, 2026.
  • Head, K.R. "Minds in Crisis: How the AI Revolution is Impacting Mental Health." Journal of Mental Health & Clinical Psychology, 5 September 2025.
  • SmarterArticles. "The Compulsive Mind: Inside the Emerging Crisis of AI-Use Addiction." 3 December 2025.