The X record

Synthetic and manipulated media

X's help pages block automated readers, so every X Help Center and Terms quote in this record was checked by hand against the live page in a browser on 2026-10-27. Labels note a policy's own date where the page shows one.

• The one-line X Rules entry bans sharing synthetic or manipulated media deceptively when it is likely to cause harm. It sits under the Rules' Authenticity heading. In the platform's own words: “You may not deceptively share synthetic or manipulated media that are likely to cause harm.”
  Source: X Help Center — The X Rules · observed 2026-10-27
• The full rule is in X's Authenticity policy, dated April 2025. It bans inauthentic, manipulated or out-of-context media when that media crosses a harm threshold. In the platform's own words: “may result in widespread confusion on public issues, impact public safety, or cause serious harm”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• Fabricated or simulated depictions of real people are on the policy's "Not Allowed" list, and the entry names artificial intelligence. This is the closest X's synthetic-media rule comes to naming AI. In the platform's own words: “media depicting a real person that has been fabricated or simulated, especially through use of algorithms or broader artificial intelligence”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• The policy lists non-deceptive sharing of manipulated media under "Allowed". The rule is about deceiving people, not about whether media is synthetic. In the platform's own words: “Sharing manipulated or out-of-context media in non-deceptive ways”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• X says outright that it may not act when it cannot tell whether media is misleading. In the platform's own words: “In situations where we are unable to reliably determine if it is misleading media, we may not take action.”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• The X Rules say X may label posts that contain synthetic or manipulated media. The synthetic-media part of the Authenticity policy does not use the word "label"; its enforcement list covers all Authenticity violations rather than media specifically. In the platform's own words: “we may label posts containing synthetic and manipulated media to help people understand their authenticity”
  Source: X Help Center — The X Rules · observed 2026-10-27

AI personas, bots and account labels

• X's Authenticity policy bans manufactured identities used for disruptive or deceptive behavior. Its "Fake personas" entry lists AI-generated profile photos among the examples. In the platform's own words: “using stock, stolen or AI-generated profile photos”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• X has an "automated" profile label for bot accounts, which the bot's operator switches on. X's profile-label page described it as currently in testing. In the platform's own words: “generating automated content not produced by a human”
  Source: X Help Center — About profile labels and checkmarks on X · observed 2026-10-27
• Labeled automated accounts must stay linked to the human-run account that manages them, so viewers can see who is behind the bot. The label became available to accounts that accepted an invitation to a test group. In the platform's own words: “Our Automation rules require these accounts to display labels and remain connected to a human-run account.”
  Source: X Help Center — About automated account labels · observed 2026-10-27
• X's Developer Policy requires anyone running a bot account through the X API to say clearly what the account is and who is responsible for it. It suggests a statement in the bio. In the platform's own words: “You should never mislead or confuse people about whether your account is or is not a bot.”
  Source: X Developer Platform — X Developer Policy (read live) · observed 2026-10-27
• Parody, fan and commentary (PCF) labels are chosen by the account holder, not applied by X. Accounts using them still have to follow the Authenticity policy. In the platform's own words: “These accounts must still adhere to our Authenticity policy.”
  Source: X Help Center — About profile labels and checkmarks on X · observed 2026-10-27

Non-consensual intimate imagery

X rewrote this policy in August 2026. The items below quote the current page; the earlier version, dated December 2021, said less about AI (see the change log).

• The current rule covers creating and manipulating intimate content of a real adult without consent, not only sharing it. In the platform's own words: “You may not share, create, manipulate, distribute or threaten to distribute content that sexualizes or exposes an identifiable real adult without their explicit consent.”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• The policy names AI directly among the ways such content can be made. In the platform's own words: “Manipulation or edits, including using Artificial Intelligence (“AI”) or any other tool”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• AI-generated and deepfake media of a specific person are listed as violations. In the platform's own words: “AI-generated, deepfake, or digitally manipulated media that sexualizes, modifies or exposes a specific individual”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• Even prompting or trying to generate such content of a real person with AI counts as a violation, whether or not it works. In the platform's own words: “Prompting, attempting to generate, generating or sharing non-consensual intimate content of real people using AI or other tools (even if unsuccessful)”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• X says it may suspend the account that first posted such media. The December 2021 version said X would suspend that account immediately and permanently; the current page says 'may'. For others who share it, X requires removal and a temporary lock, and a repeat after a warning means permanent suspension. In the platform's own words: “We may suspend any account that we identify as the original poster of intimate media that was created or shared without consent.”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• Anyone can report this content, not only the person shown. In the platform's own words: “To reduce the burden on those affected, anyone can report this type of content.”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• Where X can't tell whether the person shown consented, and the content isn't explicit, it may need to hear from that person, and limits the content's visibility meanwhile. In the platform's own words: “In those cases, we will limit visibility of the content until we hear from the victim.”
  Source: X Help Center — Non-consensual nudity policy, dated August 2026 · observed 2026-10-27
• X allows consensually produced adult content if it is labeled and not prominently displayed, and that permission explicitly covers AI-generated content. The same page says adult content must still follow the rules that lead to removal, including the non-consensual nudity policy. In the platform's own words: “This also applies to AI-generated, photographic or animated content such as cartoons, hentai, or anime.”
  Source: X Help Center — Adult Content Policy, dated May 2024 · observed 2026-10-27

Impersonation and parody

• X bans impersonating people, groups or organizations in order to deceive. In the platform's own words: “You may not impersonate other identities of individuals, groups, or organizations to deceive others.”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• Parody, commentary and fan accounts are allowed if they meet all of X's requirements: a PCF label; an avatar that is not identical to the person or entity's; and a word such as "parody," "fake," "fan" or "commentary" at the start of the account name and in the bio. In the platform's own words: “do not use an identical avatar as the entity”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27

Verification and checkmarks

• On X, the blue checkmark is tied to a paid Premium subscription plus eligibility requirements. It is not an identity or notability check. In the platform's own words: “The blue checkmark means that the account has an active subscription to X Premium and meets our eligibility requirements.”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• X's own checkmark page says the blue checkmark does not mean the account's identity was checked. X's ID verification is described separately. An archived copy of the earlier version of this page, from March 30, 2026, had no such statement. In the platform's own words: “It does not mean that the account has been ID verified.”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• The same page says X no longer calls blue-checkmark users "verified" in the EEA and now calls them "Premium" users. X attributes the change to a European Commission decision of 5 December 2025 and says it is appealing that decision in court. In the platform's own words: “misled users and breached the European Digital Services Act”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• The eligibility list requires a display name and profile photo, activity in the past 30 days, a confirmed phone number, and no recent changes to the profile photo, display name or handle. It also requires no signs of deception. In the platform's own words: “Your account must have no signs of being misleading or deceptive”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• X can take the checkmark away at any time without notice, including from accounts found to break the X Rules. In the platform's own words: “X may remove the checkmark of an account at any time without notice.”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• X says it does not endorse or validate the views of blue-checkmark accounts. In the platform's own words: “does not endorse or validate the views expressed on accounts featuring blue checkmarks”
  Source: X Help Center — How to get the blue checkmark on X · observed 2026-10-27
• A gold checkmark marks an official organization account bought through Premium Business. In the platform's own words: “The gold checkmark indicates that the account is an official organization account through Premium Business.”
  Source: X Help Center — About profile labels and checkmarks on X · observed 2026-10-27
• A grey checkmark marks a government or multilateral organization, or an official. In the platform's own words: “The grey checkmark indicates that an account represents a government/multilateral organization or official.”
  Source: X Help Center — About profile labels and checkmarks on X · observed 2026-10-27
• Identity verification is required to earn money from X's creator programs. It is listed among the eligibility standards for creator monetization. In the platform's own words: “You must complete identity verification.”
  Source: X Help Center — X's Creator Monetization Standards · observed 2026-10-27

Community Notes

• Community Notes are written by contributors and appear only when enough raters from different points of view find a note helpful. In the platform's own words: “if enough contributors from different points of view rate that note as helpful”
  Source: X Help Center — About Community Notes on X · observed 2026-10-27
• X says its own teams cannot edit Community Notes and that notes do not represent X's viewpoint. In the platform's own words: “Community Notes do not represent X’s viewpoint and cannot be edited or modified by our teams.”
  Source: X Help Center — About Community Notes on X · observed 2026-10-27
• Notes can be attached to an image or video rather than to a single post, and then appear on every post X's system matches to the same media. That is how one note can follow a re-shared misleading image. Neither this guide nor the Community Notes help page names AI-generated or manipulated media specifically; notes work as a general context tool. In the platform's own words: “Tagging notes as "about the image" (or video, or link) makes them visible on all posts that our system identifies as containing a matching image, video or link.”
  Source: X Community Notes Guide — Notes on Media & Links (official source repository, twitter/communitynotes on GitHub, read live) · observed 2026-10-27

Terms of Service

• X's current Terms of Service took effect on October 9, 2026. The AI clauses quoted below are in this version. In the platform's own words: “Effective: October 9, 2026”
  Source: X — Terms of Service · observed 2026-10-27
• The Terms make users responsible for the content they provide, create or post, and spell out that this includes inputs, prompts and outputs. In the platform's own words: “including any inputs, prompts, outputs, and/or information obtained or created through the Services”
  Source: X — Terms of Service · observed 2026-10-27
• The license users grant X over their content includes using it with, and to train, X's machine learning and AI models. In the platform's own words: “for use with and training of our machine learning and artificial intelligence models”
  Source: X — Terms of Service · observed 2026-10-27
• Under "Misuse of the Services," the Terms ban trying to get around X's safety or security controls, naming jailbreaking and prompt engineering or injection as examples. In the platform's own words: “prompt engineering or injection”
  Source: X — Terms of Service · observed 2026-10-27

Enforcement

• The Authenticity policy's "How we enforce" list covers fake personas, impersonation and misleading media. It names anti-spam challenges, denylisting URLs, Restricting Reach, temporary loss of access to features, Profile Modifications and suspension. Severe cases mean permanent suspension straight away. In the platform's own words: “For severe violations, accounts will be permanently suspended at first detection.”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• For a first or isolated offense under the Authenticity policy, X describes lighter steps. In the platform's own words: “we may take a number of actions ranging from requiring deletion of one or more posts to temporarily locking account(s).”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• "Restricting Reach" means a post stays up but is shown less. X says it may keep such posts out of search, trends and recommendations. In the platform's own words: “This may include excluding posts from search results, trends, and recommended notifications”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27
• Among its post-level options, X's general enforcement page lists adding a context label to a post. The page does not mention synthetic or manipulated media specifically. In the platform's own words: “we may add a label to the content to provide context and additional information to users.”
  Source: X Help Center — Our range of enforcement options · observed 2026-10-27
• At the account level, the same page ties suspension to repeated policy violations. In the platform's own words: “We take action to suspend an account if we determine that a user has engaged in repeated violations of our policies”
  Source: X Help Center — Our range of enforcement options · observed 2026-10-27
• Users can appeal enforcement under the Authenticity policy. In the platform's own words: “If you believe we made a mistake, you can submit an appeal.”
  Source: X Help Center — Authenticity policy, dated April 2025 · observed 2026-10-27

Change log

Dated policy changes we have documented — from the platform's own announcements or named press. Policies change; this log records what we have seen, when.

• 2020-02-04 — Twitter publishes its synthetic and manipulated media rule on its company blog, saying its teams would start labeling Tweets containing such media on March 5, 2020.
  Source: Twitter Blog — Building rules in public: Our approach to synthetic & manipulated media (Arquivo.pt capture, 2024-05-13)
• 2021-12 — Date shown on the earlier version of X's Non-consensual nudity policy (archived copy, October 2025). It covered intimate media shared without consent, including images that digitally put a person's face onto another person's nude body, and said X would immediately and permanently suspend the original poster.
  Source: X Help Center — Non-consensual nudity policy (Arquivo.pt capture, 2025-10-12)
• 2023-04-01 — X begins winding down its legacy verification program, according to its legacy verification page. Under that program, accounts had to be notable and active to be verified.
  Source: X Help Center — Legacy verification policy
• 2024-05 — Date shown on X's Adult Content Policy page. The policy allows consensual adult content if it is labeled, and explicitly applies that to AI-generated content.
  Source: X Help Center — Adult Content Policy
• 2025-04 — Date shown on X's Authenticity policy page. That page now holds the synthetic and manipulated media rule, the impersonation and parody-account rules, and the fake-persona ban, which names AI-generated profile photos.
  Source: X Help Center — Authenticity policy
• 2025-12-05 — The European Commission decides that X's blue checkmark misled users and breached the Digital Services Act. This is as described on X's own help page, which says X is appealing in court and has stopped calling blue-checkmark users "verified" in the EEA.
  Source: X Help Center — How to get the blue checkmark on X
• 2026-01-15 — An earlier version of X's Terms of Service takes effect (archived copy, March 2026). It already made users responsible for inputs, prompts and outputs and licensed user content for AI training.
  Source: X — Terms of Service (Arquivo.pt capture, 2026-03-11)
• 2026-07-04 — First seen in an archive capture on this date: X's blue-checkmark help page at a new address, stating that the checkmark does not mean the account has been ID verified. The previous page, captured 2026-03-30, had no such statement. The exact date of the change is not known.
  Source: X Help Center — How to get the blue checkmark on X (Arquivo.pt capture, 2026-07-04)
• 2026-08 — X rewrites its Non-consensual nudity policy (page now dated August 2026). It now covers creating and manipulating such content, names AI-generated and deepfake media, counts prompting an AI for it as a violation even if unsuccessful, and says X 'may' suspend the original poster.
  Source: X Help Center — Non-consensual nudity policy
• 2026-10-09 — A new version of X's Terms of Service takes effect. The AI clauses quoted in this record (inputs, prompts and outputs; licensing content for AI training; the ban on prompt injection) are in it.
  Source: X — Terms of Service
• 2026-10-27 — Record opened. Policies as observed on this date: X's help pages and Terms were checked against the live pages in a browser, and its developer policy and Community Notes guide were read live.