Why 'Just a Tool' AI Can Still Feed OCD Compulsions

AI chatbots, though often dismissed as 'just a tool,' can reinforce OCD compulsions by providing endless reassurance and eliminating social friction, according to clinicians at the Division of Digital Psychiatry at BIDMC. Users on Reddit report that AI-driven reassurance seeking worsens their OCD cycles, prompting experts to urge clinicians to assess not only which AI patients use but also how often and why.

OCD /us/basics/ocd Why 'Just a Tool' AI Can Still Feed OCD Compulsions Always-available answers can reinforce compulsions, even from 'just a tool.' Posted July 6, 2026 Reviewed by Kaja Perina /us/docs/editorial-process Key points - AI doesn't have to be relational to become clinically significant. - For those with OCD related conditions, chatbots may offer reassurance that strengthens maladaptive coping - Clinicians should assess not only which AI patients use, but also how often they use it, and why Co-authored by Jill Noorily , Mason Granof , and Kathryn Ledley at the Division of Digital Psychiatry at BIDMC. This post is part 3 of a series. Looking at the r/ OCD https://www.psychologytoday.com/us/basics/ocd Reddit board, users describe similar stories... I have been using AI for reassurance. I have health OCD and am prone to googling symptoms to ease the fear. With the chatbot introduction, it was an easier way to get constant feedback... It has become an endless source of reassurance, feeding my cycle of checking. It is wonderful in calming me down, but I keep returning and relying on it. I don't want anyone else to start this, as now it is a struggle to stop using AI for this. I need to sit with the fear, but this is so much easier to do. Many commenters echoed the same experience, with some urging others in the OCD community to avoid using AI altogether after describing how easily it became a source of compulsive reassurance and checking. Reassurance seeking has long been recognized as a mental compulsion featured in many OCD cases. Patients may repeatedly ask loved ones for confirmation, search for symptoms or situations online, or repeatedly ask clinicians for it. The relief found in reassurance is real but temporary, and each successful attempt to reduce uncertainty and anxiety https://www.psychologytoday.com/us/basics/anxiety ultimately reinforces the cycle. AI accelerates the mechanics of this cycle because, unlike human sources of reassurance, chatbots https://www.psychologytoday.com/us/basics/artificial-intelligence are always available, supportive, and generate endless variations, always continuing the conversation. AI chatbots provide people with a space to ruminate on their obsessions without being accompanied by social friction that might act as a barrier to reassurance-seeking. Can AI become part of a disorder even when it isn't acting like a companion? During the Society of Digital Psychiatry's AI Clinical Learning Collaborative, we discussed a hypothetical case like this. From this topic, the group explored the dangers of dismissing AI as “just a tool”. The group agreed that even when AI is treated strictly as a tool, it can still bring harm to the user, depending on the user’s psychiatric https://www.psychologytoday.com/us/basics/psychiatry condition. In the case of OCD, the reassurance aspect of AI adds fuel to maladaptive coping. This is why discussions about AI safety cannot stop at factual accuracy. The same technically correct response can either reinforce or interrupt an unhealthy behavioral pattern, depending on how and when it is delivered. Can AI ever truly function as "just a tool"? One participant raised another question: even in these cases, can conversational AI ever truly function as "just a tool"? Large language models are designed around conversation, and users typically interact with them through natural language dialogue. Combined with our tendency to anthropomorphize https://www.psychologytoday.com/us/basics/anthropomorphism , this makes it remarkably easy for functional interactions to take on a social quality even when users don't have that intent. For clinicians, this means that repeatedly reminding patients that "it's just AI" may do little to change how the interaction functions psychologically. Clinical insights that came from our conversation: Look beyond AI companionship. AI doesn't have to become a companion to become clinically significant. Sometimes it becomes part of a patient's illness simply by becoming an exceptionally effective tool. Ask about the dose, not just the product. The frequency of AI use is an indicator of overreliance. Ask when, how often, and why they reach for it. “Privacy” and lack of social friction can lead to overuse. For reassurance seeking and rumination, AI offers a judgment-free source of feedback that removes the social friction that might otherwise interrupt compulsive behavior https://www.psychologytoday.com/us/basics/compulsive-behaviors . Approach these conversations without shame https://www.psychologytoday.com/us/basics/shame and through a harm reduction https://www.psychologytoday.com/us/basics/harm-reduction lens. Pay Repeatedly asking the same question in different ways may reveal reassurance seeking, intolerance of uncertainty, or rumination. attention https://www.psychologytoday.com/us/basics/attention to repeated questions. References To make these discussions available beyond our collaborative, the Division of Digital Psychiatry and the Society of Digital Psychiatry created a free, interactive training at ai-digital-literacy.net http://ai-digital-literacy.net .