BotBlind
How to hide AI and bot replies on X
Why X replies fill up with AI and bots
Under almost any popular post you will find the same pattern: replies that arrive within seconds, praise the post in generic terms ("Great point! This is a game-changer 🚀"), and could have been written for any other post. Some are fully automated accounts, some are people pasting AI-generated text to farm visibility, followers or clicks, and some are plain spam pointing to a bio link, a DM or a crypto offer. They push real conversation further down the thread.
How to spot an AI or bot reply
No single sign is proof, and real people write short generic replies too. But several of these together are a strong hint:
- Template AI phrasing: "great point", "couldn't agree more", "this resonates", "a testament to", "delve", "in today's fast-paced…".
- Rhetorical templates: "It isn't X, it's Y. So the real question: …", a one-line validation followed by "Curious what…", or "Here's the thing:". People write these too, so on their own they are only a weak hint.
- Generic praise that ignores the post: the reply shares no real words or ideas with what it is replying to.
- Speed: a long, polished reply posted within a minute of the original post.
- Style tells: em dashes (—) in a short reply, and stacks of template emojis like 🚀🔥💯.
- Repetition: several replies in the same thread with near-identical text or the same three-word phrase.
- Handle patterns: usernames that end in a long string of digits.
- Spam hooks: "check my bio", "DM me", giveaways, airdrops, "100x", shortened links.
Native X settings that help
Menu names change from time to time; these are the current paths in X on the web. All of them only affect your own view or your own posts.
1. Mute words and phrases
Go to Settings and privacy → Privacy and safety → Mute and block → Muted words and add phrases that bots overuse, such as check my bio, game-changer or DM me. You can choose where the mute applies (for example Home timeline and notifications), whether it applies to everyone or only to people you don't follow, and for how long.
Limit: it is a fixed word list. Bots rephrase, and broad words also catch real people.
2. Mute or block the account
On any reply, open the ··· menu and choose Mute (you stop seeing them, they aren't told) or Block (they can't reply to or interact with your posts). Use Report post → Spam for clear spam.
Limit: one account at a time, and new bot accounts appear constantly.
3. Hide replies on your own posts
On a reply to one of your posts, open ··· → Hide reply. The reply moves out of the main conversation into a hidden replies view. It isn't deleted and anyone can still open it, and the author isn't notified.
Limit: only works on your own posts, and only reply by reply.
4. Limit who can reply
Before posting, click Everyone can reply under the composer and pick Accounts you follow, Verified accounts or Only accounts you mention. You can change it later from the post's ··· menu (Change who can reply).
Limit: it also blocks real people you don't follow, and it doesn't help on other people's threads.
5. Filter your notifications
In Settings and privacy → Notifications → Filters, keep the Quality filter on and use Muted notifications to mute notifications from accounts such as new accounts, accounts with a default profile photo, or accounts you don't follow.
Limit: it only cleans your notifications tab, not the replies you read under posts.
Native tools compared
| Method | What it catches | Works on other people's threads? | Effort |
|---|---|---|---|
| Muted words | Replies containing words you listed | Partly (where the mute applies) | Ongoing list upkeep |
| Mute / block | One account | Yes, for that account | One by one |
| Hide reply | One reply | No, your posts only | One by one |
| Who can reply | Everyone outside the group you pick | No, your posts only | Per post |
| Notification filters | Low-quality notifications | No, notifications only | Set once |
| BotBlind (extension) | Replies that score as likely AI or bot | Yes, every conversation you open | Set once, tune strictness |
Filter AI and bot replies automatically with BotBlind
BotBlind is a free Chrome extension that does the spotting for you on every X conversation page you open (x.com and twitter.com):
- Reads the replies already loaded in your browser: text, handle and timestamp. No X login, no X API.
- Scores each reply from 0 to 100 with transparent rules based on the signs listed above (AI-cliché phrases in English and Spanish, generic praise, replies unrelated to the post, em dashes and template emojis, near-duplicate or repeated phrases in the thread, digit-heavy handles, very fast replies, and promo/crypto/DM spam).
- Collapses replies above your strictness level (Relaxed 70, Balanced 50, Strict 35) into a one-line bar that shows the score and the reasons, with a Show button.
You can mark any account as ✓ Human or ⚑ Bot; the correction is saved only in your browser. BotBlind uses no AI service and makes no network requests: everything runs locally, it never deletes, reports or blocks anything, and it collects no data (see the privacy policy).
Status: version 1.0 is built and its Chrome Web Store listing is pending. It's a rules-based filter, so it will sometimes hide a real reply or miss a bot, which is why every hidden reply is one click away.
Join the free waitlist and we'll send one email when you can install it.
Join the waitlist
FAQ
Can X hide AI replies automatically?
X has no setting that specifically detects AI-written replies. Its tools work by word (muted words), by account (mute, block), on your own posts (Hide reply, who can reply) or on notifications (quality filter, muted notifications). To filter AI and bot replies under any post you read, you need a browser extension such as BotBlind.
Do hidden or muted replies disappear for everyone?
No. Muting only changes your own view. A reply you hide on your own post is moved behind a hidden replies view that others can still open. BotBlind also only changes your own view and never deletes, reports or blocks anything.
Can a detector tell AI from human replies with certainty?
No. Short replies are hard to classify and humans also write generic replies. Good filters collapse instead of deleting, show why a reply was flagged and let you undo it with one click.