X spent 2026 deleting reply bots. A timeline for people who reply by hand
Five dates that changed how replies work on X (Twitter) this year, and what they mean if you type your own.
Short answer: In 2026 X banned paid-to-post apps (January 15), stopped self-serve API apps from replying to strangers (February 23), began suspending accounts with autonomous AI replies (May 28) and removed 42,000 accounts for automated replies (July 25). Hand-typed replies are unaffected.
If your replies on X feel less crowded by "Great insight! 🚀" than a year ago, you're not imagining it. X spent most of 2026 taking automated replies apart, one rule at a time. Here is the short version, with sources.
| Date | What changed | Who it hits |
|---|---|---|
| January 15 | Paid-to-post apps banned | Apps that pay people to post |
| February 23 | API replies to strangers cut off | Auto-reply tools on self-serve API tiers |
| May 28 | Suspensions for autonomous AI replies | Accounts that let an AI reply as them |
| July 25 | 42,000 accounts removed | Automated repliers |
| September 3 | Reply filter hides 34 of 59 replies under one post | Bots, and people who write like one |
January 15: paid-to-post apps are out
X's head of product at the time, Nikita Bier, announced that apps which pay people to post would no longer be allowed, and that their API access was being revoked. The so-called "InfoFi" products, which rewarded accounts for posting about crypto projects, shut down within days.
February 23: no more replies through the API
X's developer account said that apps on the self-serve API tiers can only post a reply when the author of the original post has mentioned or quoted the replying account. In plain terms: a tool can no longer drop replies under strangers' posts for you. Scheduling your own posts was not affected.
May 28: suspensions for autonomous AI replies
X began suspending accounts that let an AI reply on its own, with no human in the loop. Using AI to proofread what you wrote stayed fine, according to PiunikaWeb's report on the change.
July 25: 42,000 accounts removed
X removed 42,000 accounts for automated replies and said it had taken down 1.7 million bot accounts since January, StartupFortune reported.
September: the filter starts catching people
By early September the filter that hides low-quality replies was hiding more than half the replies under some posts, and catching some real people whose prose read too smooth, Techweez reported. If your reply sounds like a press release, the filter can't tell you from a bot. Neither can readers.
The safest reply on X in 2026 is one a human obviously typed.
What this means if you reply by hand
- The field is emptier. Fewer bots under big posts means a real reply has more room.
- Don't paste generated replies. Even if you press the button yourself, text that reads like a machine wrote it now gets buried.
- Be specific. A number, a screenshot, a thing that happened to you. Bots are bad at specifics. We keep a list: seven replies that aren't "Great post!".
- Tools that pick, not tools that post. Anything that posts for you is on the wrong side of these rules. A list of where to show up is not.
Each of these has its own short write-up in 2026 on X, month by month. For why a hand-typed reply under the right post is worth the effort, see what X's ranking code says about replies.
X changes its rules often. Check the linked sources before you build a routine on any of this.
Want this as a daily habit?
The Club sends a daily call sheet for growing on X (Twitter): which posts to reply to right now, which format is taking off, and the one post to write today. Fifteen minutes, then you log off.
Join the club