Frequently Asked Questions (FAQ)

SocialWatch was built by a small independent team of parents, engineers, and researchers who watched the gap between what people experience on social media every day and what the people around them understand about it grow into a real safety problem. We are not affiliated with, endorsed by, or operated by Meta Platforms, Instagram, TikTok, X, or any social media platform. Our purpose is awareness, not surveillance: turning the opaque feed someone scrolls every day into something a parent, partner, friend, or mentor can read, recognize, and have an honest conversation about — early, before a quiet pattern becomes a crisis. In the cases that matter most, that conversation can prevent harm, and sometimes save a life.

SocialWatch analyzes the public content of the accounts a person follows on social media and turns that into a Network Influence Environment Assessment — a readable picture of the digital world that surrounds them. It is a research and conversation tool: a starting point for an adult who cares about someone to understand the kind of content shaping that person's day-to-day attention. It is never a substitute for clinical, legal, or professional judgment, and the report describes the environment around a person — not the person themselves.

SocialWatch is built for personal awareness: parents and guardians, partners, friends, mentors, journalists, researchers, and any adult who wants context on the public side of someone's social media presence before having a conversation or making a personal judgment. It is NOT a background check, a consumer report, or an automated decision-making system. The report must NOT be used — by you or by anyone you share it with — to make decisions about employment, hiring, firing, promotion, tenancy, credit, insurance, immigration, child custody, criminal justice, or any other decision that produces a legal or similarly significant effect on the subject. We are not a Consumer Reporting Agency under the U.S. Fair Credit Reporting Act (FCRA), and our reports are not 'consumer reports' as defined by the FCRA. If you need a report for any of those decisioning purposes, you must use a properly regulated background-check provider instead. Using SocialWatch for any of those purposes violates our Terms of Service.

We only read information that the account owner has chosen to make public. That means: public posts, public captions, public profile bios, public follow lists, and links shared in public posts. We do NOT access private accounts, direct messages, stories that aren't public, drafts, saved items, location history, app analytics, login activity, contacts, camera roll, or anything behind a login wall. We do not store anyone's password, we do not log into anyone's account, and we do not act as anyone on any platform. We do not have or claim any special API relationship with Meta, TikTok, X, or any other platform — everything we read is content that a logged-out visitor could see in their browser. We also do not perform facial recognition, biometric inference, or photo-based identification. The analysis is based on text and metadata only.

Our classification pipeline uses leading commercial large language models accessed via their official APIs. For every public account in the network, the pipeline does three things: 1. It collects a sample of the account's public posts and profile metadata. 2. The model reads that sample and assigns tags from a fixed, human-curated taxonomy (the Feed Fingerprint taxonomy). It is never asked to diagnose, profile, or pass judgment on any person — only to answer 'is this kind of content present in this public feed, yes or no?' 3. Those tags are organized around established academic and public-health frameworks (Bronfenbrenner's Social Ecological Model, the CDC's YRBSS, Search Institute Developmental Assets, the PIRUS / PERIL radicalization research, Pew Research political typology, Holland RIASEC, Schwartz Universal Values) so that the dimensions we report map to named research constructs rather than being purely proprietary categories. The mapping to these frameworks is an editorial interpretation — none of the frameworks' authors have validated or endorsed this specific application. We do not fine-tune models on user data, our model providers do not use API inputs or outputs to train their models by default (and we have not opted in to any data-sharing program), and the model only describes what is present in a public feed — it never makes a decision about a person.

Yes — your search is private. We do not log into anyone's account, we do not send a follow request, we do not interact with the profile, and the platform we read from has no way to attribute the read to the person who searched. We only ever look at content that any logged-out visitor can already see on the public web.

Yes. SocialWatch only reads content that the account owner has chosen to make public. We do not bypass any login wall, we do not access private accounts, we do not download or republish posts, and we do not store the raw content beyond what is needed to compute and explain the report. The analysis itself is an editorial and research function — comparable to a journalist or researcher reading a public profile and writing about the patterns they see. This is general information about how the platform operates, not legal advice. Laws differ by country and by use case — see the Acceptable Use section above for use cases that are explicitly out of scope.

No — the report is a starting point for awareness and conversation, not a verdict. SocialWatch is an informational tool, not a substitute for professional, clinical, legal, or medical advice. If you have serious concerns — for example about self-harm, eating disorders, hate, harassment, addiction, or radicalization — please consult a qualified professional (mental-health provider, family doctor, legal advisor, or other specialist) or contact the appropriate helpline listed under Experts. If you or someone you know is in immediate danger, contact local authorities or emergency services right away.

Treat it as a starting point for a conversation, not a conclusion. Open the per-area evidence in the report so you can read the specific public content the signal is based on, verify it for yourself, and consider the context. If the pattern looks real and concerning, the next step is a calm conversation with the person involved and, where appropriate, a qualified professional. If you or someone you know is in immediate danger, contact local authorities or emergency services right away.

AI systems can produce false positives and misread nuance, irony, and context. If a result seems wrong, open the per-area evidence for that signal, review the original public content directly, and weigh it against what you know about the person's situation. If the discrepancy is significant — or if you are the owner of an account you believe is being misclassified — please email contact@socialwatch.ai and we will review.

Yes. We never ask for your social media password and we cannot log in as you. We do not sell your data, we do not use it to train AI models, and we do not share it with advertisers. The public posts we read are kept only as long as needed to generate and explain your report, after which they are purged from working storage; the long-lived record is the derived signals (tags, scores, area aggregates) used to render the report on demand. If you are an account owner and you do not want your public profile included in current or future SocialWatch reports, email contact@socialwatch.ai with your handle and we will exclude you. Full details are in our Privacy Policy.

If you encounter technical issues, want to report an inaccurate result, request removal of an account from current and future reports, or have any other question, email contact@socialwatch.ai. We appreciate the feedback — it is the main way the classifier and the platform improve.