Paste a username, bio and recent posts — we estimate bot/fake likelihood from random usernames, promo bios and repetitive posts, with no model. A rough guide.
Free with no sign-up, and the check runs inside your browser, so what you enter never leaves your device. The result is a signal to help you decide, not proof for money or legal decisions.
Estimate whether a social account is an automated bot or a fake account used for manipulation, from the characteristics of its username, bio and posting pattern — signals invisible if you only look at follower counts.
A username with a random digit string (user_837261), or a letter-number mix unrelated to a human name, can indicate a mass-created account. An empty bio — or, conversely, one filled only with hashtags, external links and promo copy — is also a signal.
Posting patterns matter too. Repeating the same phrase at short intervals, dropping links with no conversation, or activity locked to mechanically regular time windows all raise the odds of automation.
Signals from three axes — username, bio and posts — are combined into a "bot/fake likelihood" score. The more you fill in (all three fields), the steadier the read.
Do not conclude from a single signal. New accounts, or genuine promotional accounts, can show similar signs. Suspicion grows when several signals overlap.
Bot is not automatically a slur. Earthquake alerts, release notifications and repository change feeds are automated accounts worth following. Posts written by a person and published through a scheduling tool look mechanically spaced while the content is entirely human.
What actually needs catching, then, is deception rather than automation. Does the automated account say what it is, or does it pass as a person in order to move opinion or money? Major platforms draw their policy lines the same way — concealed manipulation is prohibited, automation as such is not.
Read the score in that order. If signals come back high, look first at what the account says it is, then at whether its behaviour matches that description. Where there is no mismatch, an automated account is not a problem.
Accounts used for manipulation look unremarkable one at a time. The real signal lives between them: the same sentence posted from several accounts within minutes, dense mutual following with almost no connections outward, creation dates clustered into a particular few weeks.
Profile photos give the group away too. The same image reused across accounts is findable through reverse image search — while an image that appears nowhere at all may be generated, which points to an image check instead. That two opposite results both count as grounds for suspicion is characteristic of this area.
This tool reads the text of one account. A well-organised network's individual members can therefore score low. When several accounts are in question, run them one by one and compare, and look at how alike their bios and posts are. That resemblance is stronger evidence than any individual score.
Six or seven accounts with unrelated names and photos posted near-identical sentences within a few minutes. Run individually, none of them looked odd and all scored low. This is the limit of per-account analysis in plain view. The signal is between the accounts, not inside them. Line up their bios and post wording side by side and check whether creation dates and posting times cluster. That resemblance outweighs any individual score.
The username ended in eight random digits, the bio was a single emoji, and the account posted the same link dozens of times a day with only the wording changed. Name randomness, bio boilerplate and post repetition all rise together, producing a high score. The point is that three signals fire at once — with only one lit, withhold judgement.
Forty followers, most posts are replies to other people, and posting times are scattered across the day and week. The score stays low. What matters here is not scale but uniformity — humans are irregular in timing and sentence length, automation is the opposite.
Brand accounts, news curation feeds and solo marketers using scheduling tools produce exactly the bot profile: high link ratio, uniform formatting, regular cadence. The signals are not wrong — those accounts really are partly automated. The question this tool answers is "is this pattern automated?", not "is a human behind it?". Before reporting or blocking, look for things only a person does: replies, corrections, answering questions.
Upload the profile photo to the image detector on the home page to check for AI generation. This tool handles text signals.
No. Early on there is simply little information and few signals. It is more accurate to re-check once an activity pattern builds up.
Do not call it out publicly. Naming a real person's account by mistake does real damage. Use the platform's reporting tools and mute or block to reduce your own exposure.
Not directly. This reads the text an account writes, not its follower list. But accounts with bought followers tend to show engagement far below what their follower count implies, so check that ratio yourself alongside the score.