X reply limits in 2026: what the API restriction means for AI reply tools

On February 23, 2026, X restricted replies posted through its API. An app on self-serve API access can now reply to a post only if the post's author @mentioned the replying account or quoted one of its posts. The change applies to apps and bots that reply through the API. It doesn't change what happens when you type or paste text into X's own reply box and press Reply.

Disclosure: I build Voxily, a Chrome extension that drafts replies inside X's reply box. You press Reply yourself, and Voxily doesn't post through the X API.

What X changed

X announced the change in its developer forum in a post titled X API v2 Update: Addressing LLM-Generated Spam, and the @XDevelopers account posted a summary the same day. X wrote that programmatic replies had been heavily used to insert low-quality or automated content into conversations, including by AI-assisted developers.

Before the change, any authenticated app could reply to any post through the API. Now X accepts an API reply only when the original author has "summoned" the replying account: the author @mentions that account in the post, or quotes one of its posts. If neither is true, X rejects the reply.

The same update limited @mentions and quotes sent through the API. An app could only @mention or quote people who were already part of the conversation. X's changelog entry of April 16, 2026 goes further: it says quote posts through the API have been removed from all self-serve tiers. Posts that aren't replies can still be created through the API within those limits.

The rule is still in X's documentation. The February 23 entry in the X API changelog describes it, and the X API integration guide for posts lists it under "Post creation restrictions (self-serve)".

Who it affects

The restriction covers replies created through the X API on self-serve access. X's announcement lists the legacy Free, Basic and Pro subscriptions and pay-per-use packages. Enterprise access isn't affected. For developers who need to reply without being mentioned, such as for customer service at scale, X suggests Enterprise access.

In practice, this affects reply bots and any app that answers posts on its own through the API. In a thread on X's developer forum, people running bot accounts labeled as automated asked whether X would exempt them. One said the change would hurt the engagement their bot had built.

What it doesn't change

The restriction is about how a reply reaches X, not about who wrote the words. When you type or paste text into the reply box on x.com or in the X app and press Reply, you aren't sending it through the X API that the restriction covers. X's announcement is about replies made through the API and doesn't mention replies people post themselves.

The same is true when the text started as an AI draft. The restriction applies to replies sent through the API, whoever or whatever wrote them. A draft you review, edit and post from the reply box is a reply you posted.

Having a browser tool press Reply for you isn't a way around the change either. X's automation rules tell people not to use non-API automation, such as scripting the X website, and say it may lead to permanent suspension.

X's rules still apply to your replies

Posting from the reply box puts you under the same rules as anyone else on X. X's help page on limits lists 50 original posts and 200 replies per day for unverified accounts. It says those limits count actions from all devices, including web, mobile and the API.

The X Rules prohibit using X in a way intended to artificially amplify or suppress information, or to manipulate or disrupt people's experience on X. The automation rules add that the reply and mention features are meant to make communication between people easier. Automating them to reach many users without being asked isn't permitted, and X gives replying automatically to posts found by keyword searches alone as an example.

The automation page also says you are responsible for the actions taken with your account, including by apps you connect to it. A reply drafted with AI and posted by you is your reply. If it's spammy or repetitive, X's rules apply to it as they would to any reply you typed.

Voxily's privacy and platform rules page lists what the extension processes when you ask for a draft and what it doesn't do.

Why fewer, specific replies beat volume

X described the API change as a response to automated, low-value replies and unsolicited engagement. AI makes that kind of reply cheap to produce, whether a bot posts it or a person does.

A useful reply answers something in the post. The author can respond to a question about the step where their signups stall. They can't do much with "Great insight, thanks for sharing." A generic reply gives the author nothing to answer, and it reads the same under any post.

When you use AI for a first draft, the edit is where the reply becomes yours. Pick one point from the post, cut praise that could fit any thread and remove any claim about your experience that isn't true. The guide on editing AI drafts for X walks through examples.

Replying less often also leaves time to read the thread before you join it. If you can't say which line of the post your reply answers, skip it. A few replies that start conversations are worth more than many that nobody answers.

Questions about X's 2026 reply restriction

Does the restriction affect replies I post myself?

No. It covers replies sent through the X API on self-serve access. Replying from the reply box on x.com or in the X app isn't covered by it. Your account's limits and X's rules still apply.

Can an AI tool reply for me through the API?

On self-serve access, only to posts that @mention your account or quote one of its posts. X's automation rules add conditions for automated replies, including that the person receiving them has opted in, and they say any AI reply bot needs X's prior written approval.

Is the limit of 200 replies a day part of this change?

They are separate. The API restriction was announced in X's developer forum. The 200 replies per day figure is on X's help page about account limits. It applies to unverified accounts and counts actions from every device.

Before your next reply, find the line in the post it answers. If you can't, don't post it.