Googling, Reddit, and AI Chat: When Online Reassurance Seeking Starts Working Like an OCD Compulsion
Updated: 37 minutes ago
Last reviewed: 09/14/2026
Reviewed by: Dr. Kiesa Kelly
You type the thought into the search bar. Six tabs later you feel steadier, for about four minutes. Then a slightly different phrasing occurs to you, and you start again.
If that is a familiar evening, you are not lazy and you are not simply being thorough. Searching symptoms, scrolling forums, and asking an AI chatbot the same question twelve ways can do what asking a reassuring person does: it takes the edge off the doubt, and it teaches the doubt to come back. This article is about the digital side of that pattern, where it is hardest to recognize as a compulsion.
In this article, you'll learn:
Where the evidence on online reassurance seeking honestly stands right now
How the pattern differs across search engines, forums, and AI chat
The single question that separates a compulsion from genuine research
Why some people go online instead of asking a person
What response prevention looks like surface by surface, and what to ask a provider
What it is: the one-paragraph answer
Online reassurance seeking is what happens when the urge to resolve an obsessive doubt gets routed through a screen instead of a person. In OCD a compulsion is a repetitive behavior or mental act performed to reduce the distress an intrusive thought creates [10][11], and excessive reassurance seeking is a long-recognized maintenance behavior in OCD [9]. If yours runs mainly through people, our companion piece on why asking for reassurance stops working and what therapy does instead covers that ground, and this piece assumes it rather than repeating it. Only the surface has changed. What makes the digital version hard to catch is that it looks like diligence.
Here is the honest state of the evidence, stated plainly and up front. No published study has yet tested whether online reassurance seeking behaves like a compulsion in OCD in the way that matters clinically: brief relief, escalation over time, and long-term maintenance of the disorder. Clinicians see the pattern in the room. The early research is consistent with it [1][2]. But the researchers working on this describe online reassurance seeking as less thoroughly studied than the interpersonal kind [1], and we would rather tell you that than imply a certainty the literature does not yet support.
🔎 Key takeaway: The behavior is well described clinically and under-tested experimentally. That is a reason to take it seriously and a reason not to overclaim.
Three things people get wrong about it
"If I'm reading real medical sources, it's research, not a compulsion." A peer-reviewed paper read forty times to settle the same doubt is a compulsion; a single search that answers a question and ends is not. Function is the variable, not source quality.
"It only counts if I lose hours to it." Duration is a poor test. Compulsions are defined by function, not by clock time, so a ninety-second check repeated through the day can matter as much as one long research session — it is the repetition that keeps pairing relief with checking, not the total minutes.
"Asking an AI is safer than asking a person, because I'm not burdening anyone." Half true. You are not burdening anyone, and the NPJ Digital Medicine model argues that this is part of the problem: a chatbot exchange lacks the ordinary social friction — fatigue, pushback, the cues that signal enough — that eventually slows a conversation with a person down [5]. That is a proposed mechanism, not a demonstrated one — and as the section below explains, the evidence does not show digital reassurance is harder to stop than asking a person, only that for some people it is more private.
What it looks like across the three surfaces
The loop is one loop wearing three costumes. The IOCDF's survey of digital reassurance traps spans search engines, medical sites, social platforms and AI tools [9]; in our clinical experience the response-prevention plan differs for each.
Search engines: symptom-checking and the research loop
Imagine your chest feels tight and the thought arrives that something is wrong with your heart. You search the symptom, the first result is reassuring, and for a moment the tightness seems less meaningful. Then you notice it did not mention your age, so you search again with your age added, and then a forum post mentions a rare presentation, so you search that. Ninety minutes later you feel worse than when you started, and every single search felt reasonable at the time.
That pattern has a name in the literature: cyberchondria. It is not a formal psychiatric diagnosis, and the ICD does not address it specifically [3]. A systematic review of 61 articles found that none examined a clinical sample with a diagnosed mental disorder, that "compulsive or repetitive behavior" appeared in only about two thirds of published definitions, and that the link to health anxiety is far stronger than the link to OCD, which in one study vanished once negative affect and health anxiety were accounted for [3]. Real, useful, and genuinely unsettled.
This matters, because the two produce an identical browser history for different reasons. In health anxiety the fear is of the illness. In OCD the searching serves the doubt rather than the topic, and the feared content can be anything. They overlap, and reassurance seeking serves a similar function in both [4], so it is worth knowing how health anxiety and OCD differ before deciding which you face.
Forums and AI chat: strangers and chatbots as the new accommodators
Now imagine a doubt too embarrassing to say out loud, a thought that feels morally repugnant and completely unlike you. You would not ask your partner, because asking would mean saying it. So at two in the morning you find a subreddit where dozens of people describe something close to yours, and the relief is enormous because you are not the only one. A week later you are back, because that thread described a thought that was almost like yours, and the gap has become the new doubt.
Forum reassurance is inexhaustible and calibrated by strangers. Because posters describe the same experience slightly differently, the comparison itself generates new doubt. This is common with the less stereotypical presentations of OCD, including harm, moral, and relationship themes, where the thought's content is what makes it too shameful to raise with anyone who knows you.
AI chat adds what forums do not: a responder that answers instantly, personally, and without limit. A 2026 paper in npj Digital Medicine proposes a transdiagnostic model in which repeated chatbot interaction perpetuates OCD and anxiety by reinforcing avoidance, intolerance of uncertainty, "need to know" compulsions, and perfectionism [5]. Be accurate about what it is: a well-argued framework, not a trial. In a 2026 letter to the Irish Journal of Psychological Medicine, two Dublin psychiatrists raised the same concern, drawing in part on the same authors' earlier work [6]. A letter is an argument, not data.
Is this a compulsion, or is it just research?
Most people asking this want to know whether the thing they do at midnight is the disorder or just them.
What an evaluation looks at
A psychological evaluation for OCD does not count your searches. It looks at the relationship between an intrusive thought, the distress it produces, and what you do next to make that distress go away. A clinician will want to know how long the relief lasts, and what happens if you do not do it. That last one is the most informative: researching stops when you are finished, and a compulsion does not stop even when you know the answer.
A structured self-report measure like the Dimensional Obsessive-Compulsive Scale gives you and a clinician a starting picture across symptom dimensions [12]. It is a screener, not a diagnostic test: it organizes the conversation rather than replacing it.
The tell: what the behavior is FOR, not how long it takes
Ask what the behavior is for. If you are searching to learn something, the question can be answered, and when it is answered you stop. If you are searching to make a feeling go away, the question keeps regenerating, the relief gets shorter each time, and stopping feels unsafe rather than unfinished.
So: if the searching ends when the information arrives, it is research. If it ends only when you feel better and restarts when the feeling returns, it is a compulsion, regardless of the sources or the time it took.
🧭 Key takeaway: Research ends with an answer. A compulsion ends with a feeling, and feelings do not stay ended.
Why some people go online instead of asking someone
There is a tempting story here: that digital reassurance is stickier than human reassurance because it is always available. No study has tested that directly, and its premise does not hold up.
A 2025 study in the Journal of Anxiety Disorders compared 62 people with OCD, diagnosed by structured clinical interview, against 58 people without [1]. It found no group difference in where people preferred to seek reassurance. Seventy-three percent of the OCD group preferred asking a person, against seventy-nine percent of the comparison group — not a statistically significant difference. People with OCD were actually less likely to seek reassurance online about physical health.
What the same study did find is specific to OCD. Among those who chose online over interpersonal reassurance, people with OCD cited shame and fear of being judged significantly more often than the comparison group, and the authors point to self-concealment as a theme that may be genuinely distinctive. An earlier study in non-clinical samples found the same constellation, linking online reassurance seeking to shame and fear of one's own character [2], though that work was correlational and did not involve diagnosed OCD.
Clinically, this fits what we see. The doubts that drive people online are the unspeakable ones: thoughts about harm, about sexuality, about whether you are secretly a bad person. The search bar is chosen not because it is better, but because it does not have a face. The compulsion nobody knows about is the one that never reaches treatment. If shame is why yours has stayed private, the clinicians who do this work hear these thoughts routinely and do not find them shocking.
🤐 Key takeaway: The evidence does not say digital reassurance is more compelling than human reassurance — only that for some people it is more private, and privacy is what keeps it going.
What actually helps
ERP and response prevention, surface by surface
Exposure and response prevention has the strongest evidence behind it for OCD, and the principle is simple: make contact with the doubt, and do not perform the behavior that neutralizes it. NICE guidance is explicit that for people with obsessive thoughts, treatment should include exposure to those thoughts plus response prevention of mental rituals and neutralizing strategies [7]. A search, a forum scroll, and a chatbot question are all neutralising strategies. If the approach is new, our introduction to how ERP works is the place to start.
Write a separate plan for each surface.
Search engines. The target is the second search, not the first: most plans allow the initial question and prevent the re-check, the re-phrase, and the second-source confirmation.
Forums. The target is comparison. Reading one thread and closing the tab differs from reading until someone's description matches yours exactly.
AI chat. The target is the follow-up. Because a chatbot always produces another answer, the plan usually needs a hard rule about the number of exchanges.
NICE also addresses something readers miss: when family or partners have become involved in compulsive behaviors, avoidance, or reassurance seeking, the plan should help them reduce that involvement [7]. If someone else is doing the googling for you, that is still accommodation. Structured therapy for OCD builds these plans deliberately rather than leaving them to willpower.
Why "just one more search" resets the clock
The mechanism is the one Salkovskis described for safety behaviors, and reassurance seeking in OCD fits closely: the relief is temporary, and the behavior prevents the feared consequence from being disconfirmed [8]. You never find out the doubt would have faded on its own, because you always intervened first.
The learning your brain needs is not "the answer was reassuring." It is "I did not check, and the feeling passed anyway." Every check overwrites that lesson.
You may have met the fifteen-minute rule here. It comes from psychiatrist Jeffrey Schwartz's 1996 book Brain Lock, inside the Refocus step of his four-step method, and Schwartz is clear it is not passive waiting but an interval spent actively engaged in something else. Two honest caveats: it is not ERP and appears in neither the NICE nor the APA guideline, and the number fifteen is a clinical convention rather than a figure any trial produced. As practice at delay it is reasonable; as a countdown to a search you have already decided to make, it is another ritual. A clinician in specialized therapy turns a delay tactic into a response-prevention plan.
⏳ Key takeaway: Response prevention is not about resisting forever. It is about not intervening long enough to find out what the feeling does on its own.
When to get evaluated
Talk to a clinician when the searching starts costing you something you would rather keep: sleep, work, presence with people, or the ability to sit with ordinary uncertainty. You do not need a diagnosis in mind first, and it does not need to be severe to justify asking.
Four questions worth asking a provider before you book:
Scope. Do you treat OCD specifically, including intrusive thoughts and mental compulsions, or anxiety more broadly?
Methodology. What does your ERP plan look like for compulsions that happen on a phone rather than in a room?
History. How do you gather history when someone has been hiding the behavior and cannot easily say when it started?
Output. What will I leave the evaluation with, beyond a label? What does the first month of treatment actually look like?
If the answer to the second is vague, keep looking. Our OCD services page describes our approach, and it is reasonable to ask us the same four questions.
📋 Key takeaway: The threshold is not severity. It is cost. If the checking is taking something from your day, that is enough of a reason to ask.
Next step: getting support
The loop you are in is not a character flaw or a failure of discipline. It is a behavior that works brilliantly in the short term and expensively in the long term. The question this article opened with comes down to one you can ask tonight: was that search for information, or for a feeling? If it was for a feeling, the searching is part of the problem rather than the way out, and that is good news, because it is the part treatment knows how to change.
Think it might be OCD?
OCD responds well to the right approach — a clinician trained in ERP and I-CBT can help you tell OCD apart from anxiety and build a plan that fits.
Frequently Asked Questions
How do you stop reassurance-seeking OCD when it happens online?
When the searching is functioning as a compulsion — a pattern clinicians see often, though no published study has tested it directly — it responds to the same approach as any other compulsion: notice the urge, delay the behavior, and let the doubt stay unresolved long enough for your brain to learn it can be tolerated. In practice that means exposure and response prevention with a trained clinician, not willpower alone. NICE guidance names response prevention for mental rituals and neutralizing strategies, and a search or a chatbot question usually functions as one.
Does using ChatGPT or an AI chatbot for reassurance make OCD worse?
It can, though the mechanism is proposed rather than demonstrated. A 2026 perspective paper in NPJ Digital Medicine argues that a chatbot exchange lacks the ordinary social friction — fatigue, pushback, the cues that signal 'enough' — that eventually slows a conversation with a person down, and that repeated checking this way reinforces intolerance of uncertainty and 'need to know' compulsions. It is a framework paper, not a trial, so treat it as a well-argued hypothesis.
What is the 15-minute rule for OCD, and does it work on the urge to google?
The fifteen-minute rule comes from psychiatrist Jeffrey Schwartz's 1996 self-help book Brain Lock, and it asks you to actively refocus onto another activity for about fifteen minutes before deciding whether to act on an urge. It is not exposure and response prevention, it does not appear in NICE or APA guidance, and no trial established the number fifteen. Used as a way to practice delay it can help; used as a countdown to a permitted search it becomes another ritual.
What should I tell a therapist about the googling and forum reading?
Tell them the specifics, including the parts that feel embarrassing. The useful details are what triggers a search, how many times a day it happens, how long the relief lasts, and which surfaces you use, because a plan for search engines looks different from a plan for forums or a chatbot. Shame is a common reason this stays private — in one study, people with OCD who chose online reassurance cited shame and fear of judgment more often than people without OCD — and what stays private stays invisible to treatment.
About the Author
Dr. Kiesa Kelly is a licensed clinical psychologist in Tennessee whose graduate therapy training focused on OCD. Her pre-doctoral training included a cognitive behavioral therapy practicum at The Chicago Medical School's Anxiety Disorders Clinic, delivering exposure and response-prevention therapy to adult and pediatric patients with anxiety disorders including OCD, and she has since trained in inference-based CBT through the OCD Training School. She treats OCD using inference-based CBT, exposure and response prevention, and acceptance and commitment therapy. As a neuropsychologist by training she also has more than 20 years of experience with psychological assessment, and her assessment work includes ADHD and autism.
Dr. Kelly's doctoral training in clinical psychology, with a concentration in neuropsychology, was completed at Rosalind Franklin University of Medicine and Science, following an undergraduate degree in psychology and neuroscience at Bowdoin College. Her clinical training includes work at the University of Chicago, Vanderbilt University, the University of Wisconsin, and the University of Florida, and as an NIH National Research Service Award postdoctoral fellow she conducted research on dual pathway models of ADHD using high-density event-related potential recording. She is a member of the American Psychological Association, the Anxiety and Depression Association of America, the Tennessee Psychological Association, and the Association for Behavioral and Cognitive Therapies.
References
1. Parsons CA, Kim HJ, Singh S, Lkhagva T, Wang J, Alden LE. Covert or connected: motivations for online and interpersonal reassurance-seeking in OCD. Journal of Anxiety Disorders. 2025;115:103057. PMID: 40795489. https://doi.org/10.1016/j.janxdis.2025.103057
2. Parsons CA, Alden LE. Online reassurance-seeking and relationships with obsessive-compulsive symptoms, shame, and fear of self. Journal of Obsessive-Compulsive and Related Disorders. 2022;33:100714. https://doi.org/10.1016/j.jocrd.2022.100714
3. Vismara M, Caricasole V, Starcevic V, et al. Is cyberchondria a new transdiagnostic digital compulsive syndrome? A systematic review of the evidence. Comprehensive Psychiatry. 2020;99:152167. PMID: 32146315. https://pubmed.ncbi.nlm.nih.gov/32146315/
4. Halldorsson B, Salkovskis PM. Why do people with OCD and health anxiety seek reassurance excessively? An investigation of differences and similarities in function. Cognitive Therapy and Research. 2017;41(4):619-631. https://doi.org/10.1007/s10608-016-9826-5
5. Golden A, Aboujaoude E. A transdiagnostic model for how general purpose AI chatbots can perpetuate OCD and anxiety disorders [perspective]. npj Digital Medicine. 2026;9:343. PMID: 41826639. https://doi.org/10.1038/s41746-026-02531-7
6. Friars D, Flynn G. Generative AI and reassurance-seeking in OCD [letter to the editor]. Irish Journal of Psychological Medicine. 2026;43(2):236. https://doi.org/10.1017/ipm.2025.10138
7. National Institute for Health and Care Excellence. Obsessive-compulsive disorder and body dysmorphic disorder: treatment. Clinical guideline CG31. Last reviewed 11 July 2024. https://www.nice.org.uk/guidance/cg31
8. Haciomeroglu B, Karanci AN, Inozu M, et al. The role of reassurance seeking in obsessive compulsive disorder: the associations between reassurance seeking, dysfunctional beliefs, negative emotions, and obsessive-compulsive symptoms. BMC Psychiatry. 2020;20:356. https://doi.org/10.1186/s12888-020-02766-y
9. International OCD Foundation. Digital reassurance seeking in OCD. 21 July 2026. https://iocdf.org/blog/2026/07/21/digital-reassurance-seeking-in-ocd/
10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR). 2022. https://www.psychiatry.org/psychiatrists/practice/dsm
11. National Institute of Mental Health. Obsessive-compulsive disorder. https://www.nimh.nih.gov/health/topics/obsessive-compulsive-disorder-ocd
12. Abramowitz JS, Deacon BJ, Olatunji BO, et al. Assessment of obsessive-compulsive symptom dimensions: development and evaluation of the Dimensional Obsessive-Compulsive Scale. Psychological Assessment. 2010;22(1):180-198. PMID: 20230164. https://doi.org/10.1037/a0018260
Disclaimer
This article is for informational purposes only and is not a substitute for individualized clinical advice, diagnosis, or treatment. Reading about a pattern is not the same as being assessed for it. If you are concerned about obsessive thoughts or compulsive behaviors, please speak with a licensed mental health professional. If you are in crisis or thinking about harming yourself, call or text 988 in the United States to reach the Suicide and Crisis Lifeline.