X Made with AI Labels: What X Requires in 2026
X marks some AI-generated media with Made with AI labels, and since March 3, 2026 it has tied one disclosure rule to creator payouts. Creators who post AI-generated videos of an armed conflict without a Made with AI disclosure lose payouts for 90 days. Two older X policies also apply to AI posts: the authenticity policy on synthetic media and the automation rules. If you run an agent or tool that posts to X, this covers what X requires today and how its labels line up with the EU AI Act.
What is a Made with AI label on X, in one sentence?
A Made with AI label is an indicator X attaches to a post to tell viewers that its media was generated or altered with AI.
X describes the labels in its Media Literacy Action Plan, dated July 2026. The plan says X expanded its labels for manipulated and AI-generated media, including Made with AI indicators. The stated goal is to help EU users identify synthetic content. It sits alongside X's older labels for sensitive content and for posts with restricted reach.
How does a Made with AI label differ from a misleading media action?
X runs two separate systems: a disclosure label for AI media, and enforcement against media that deceives.
| Made with AI disclosure | Synthetic and manipulated media policy | |
|---|---|---|
| Purpose | Tell viewers media is AI-made | Act on media that misleads and can cause harm |
| Covers | AI-generated or AI-altered media | Deceptively altered or out-of-context media |
| Applies to honestly shared AI media | Yes, it is a label | No, sharing it non-deceptively is allowed |
| Consequence of skipping it | 90-day payout suspension for unlabeled AI war videos | Reduced reach, feature limits or suspension |
| Where X documents it | Media Literacy Action Plan, creator program pages | Authenticity policy |
Sources: X's authenticity policy and Original Content Rewards page, plus the Media Literacy Action Plan.
For an agent that posts media, a labeled AI image breaks no rule. An unlabeled fake of a real person shared as authentic can draw enforcement.
When does X require you to disclose AI content?
Since March 3, 2026, one rule has carried a hard penalty. Post an AI-generated video of an armed conflict without the Made with AI disclosure, and X suspends your creator payouts for 90 days. Later violations lead to permanent suspension of payments. X says it may adjust the policy as the conflict situation evolves.
The rule first appeared under Creator Revenue Sharing. That program sunset on September 7, 2026, and X now points creators to Original Content Rewards. The Original Content Rewards page carries the same armed-conflict rule. The penalty only touches payouts, so it reaches creators in the program. The page describes no penalty for accounts outside it.
Beyond that rule, the help center pages we read describe labels as a transparency tool. They do not set a general obligation to label every AI post. They also do not say how to add the disclosure. Check the post composer in your own account.
What does X's synthetic media policy prohibit?
The authenticity policy, dated April 2025, bans misleading media. That means media that may cause widespread confusion on public issues or serious harm. Two kinds are named:
- Manipulated media shared as authentic, including media depicting a real person that has been fabricated or simulated, especially with AI.
- Out-of-context media shared to deceive, such as real footage given a false location or date.
Sharing manipulated or out-of-context media in non-deceptive ways is allowed. X says it may use its own technology or reports from third-party partners to judge whether media is manipulated. Where it cannot reliably tell, it may take no action. Enforcement depends on severity and history. It ranges from restricting a post's reach to suspending the account.
What do X's rules mean for AI agents that post?
X updated its automation rules in April 2026. They apply to accounts and apps that act without a person performing each action. For agent builders, the rules say:
- You are responsible for what applications do with your account. Violations can get posts filtered from search or the account suspended.
- Deploying or operating an AI reply bot requires prior written and explicit approval from X.
- Automated replies and Direct Messages need the recipient to have opted in first.
An agent that generates and posts images for you needs X's written approval before it replies automatically. It must also label AI media where X requires it and avoid posting anything that reads as a deceptive fake.
How do X's labels fit the EU AI Act?
The EU AI Act puts transparency duties on both ends. Article 50 requires providers of systems that generate synthetic content to mark outputs in a machine-readable format. It requires deployers that publish deep fakes to disclose that the content is artificially generated or manipulated. Under Article 113, the regulation applies from August 2, 2026.
X's own plan ties its Made with AI indicators to users in the EU. One open standard for marking files is C2PA Content Credentials. An agent that produces media for EU audiences may need both: the Made with AI label on X, and machine-readable marking inside the file under Article 50.
The bottom line
If you or your agents are in X's payout program, disclose every AI video of a war. Skipping it costs 90 days of payouts. Get written approval before running an AI reply bot. Keep a record of which posts carried AI media, and keep provenance metadata in files you publish to EU audiences.
How does an agent remember which content it generated and how to disclose it?
Disclosure rules are easy to forget between sessions. An agent that drafts posts on Monday and publishes them on Thursday has to know which images it generated and which platform rules apply. The agent has to apply the armed-conflict rule and the reply-bot approval rule on every post. A coding or content agent starts each session without that record unless someone writes it down.
memU gives the agent a place to write it down as readable Markdown. A scheduled background task slices new session logs into jobs. The agent distills each job into memory and skill Markdown, such as a "publish to X" checklist. memU commits the result and makes no LLM or chat calls of its own. It stores, embeds and retrieves the Markdown the agent prepared. A standing instruction in CLAUDE.md or AGENTS.md tells the agent to retrieve before answering. All hosts share one backend, so the checklist Codex learned is available to Claude Code. It keeps a record of what the agent wrote, and labeling and rule checks stay with the agent.
The hosted version is free and syncs across devices. Get an API key from memu.so and send your agent one message:
Read https://memu.pro/SKILL.md, follow its instructions to install and configure memU, API Key is
YOUR_MEMU_API_KEY.
Self-hosting and the supported agents are listed in the memU GitHub repository.
FAQ
Does X require labels on all AI-generated posts?
Not in the help center pages we read. The one rule with a stated penalty covers AI-generated videos of an armed conflict posted by creators in X's payout program. Deceptive synthetic media is banned under the authenticity policy whether or not it carries a label.
What happens if I post an unlabeled AI war video on X?
If you are in the Original Content Rewards program, X suspends you from payouts for 90 days. Later violations lead to a permanent suspension of payments.
Can I run an AI reply bot on X?
Yes, with permission. X's automation rules say any AI reply bot needs prior written and explicit approval from X, and automated replies need the recipient to opt in.
Is AI-generated content allowed on X?
Yes. X's authenticity policy allows sharing manipulated or AI-made media in non-deceptive ways. It bans misleading media that can cause widespread confusion on public issues or serious harm.
Does the EU AI Act require AI labels?
Article 50 requires machine-readable marking of synthetic content by providers and disclosure of deep fakes by deployers. The regulation applies from August 2, 2026.