What changed in X's sensitive media policy in 2026?

X's 2026 sensitive media framework expands the category coverage, tightens auto-detection, and changes the default user experience for new accounts in the EU and several other jurisdictions. The category set now includes five families rather than the two that dominated the 2024 policy. The first family is graphic violence — content depicting injury, death, weapons in active use, or the aftermath of violent events. The second is hyper-realistic gore — wound photography, surgical content outside an educational context, and forensic imagery. The third is suggestive content that does not meet the adult content threshold but exceeds what the platform considers safe for default display — partial nudity in artistic context, intimate imagery without explicit content, and sexualised wellness or fitness content. The fourth is medical content with potential distress impact — clinical procedures, severe pathology images, and emergency medicine documentation. The fifth is the legacy adult content category, which now sits inside the Adult Content Creator program with its own three-tier classification and is separately governed. The auto-detection model has also changed. The 2026 model treats text and image jointly rather than independently, which means that a benign-looking image accompanied by suggestive text or a suggestive image accompanied by clinical text can both be reclassified. The user-side default has shifted as well. New EU and UK accounts default to 'hide sensitive media' under the platform's DSA-driven recommender system obligations, and minors' accounts cannot toggle the setting off. Advertisers face a parallel set of changes that affect ad adjacency, opt-out granularity, and the brand safety reporting that the platform now publishes quarterly. For ongoing tracking of X policy changes, see Policy Tracker .

How does X's auto-detection model decide what counts as sensitive in 2026?

X's 2026 auto-detection pipeline combines image classification, joint text-image embedding analysis, and account-level risk signals. The pipeline runs at upload time and continues running asynchronously after publication, which produces a small but material rate of retroactive reclassification — content that was originally not flagged can be flagged later if the model is re-trained or if the joint embedding score crosses a threshold. The pipeline operates in three stages. Stage one is content recognition, which classifies the image or video frame against the five sensitive media families and outputs a confidence score per category. Stage two is context fusion, which evaluates the surrounding post text, the embedded link target where applicable, the conversation thread, and any prior account behaviour to adjust the classification. Stage three is policy application, which maps the final classification to the sensitive media flag, the default visibility rule, and the ad-adjacency status. The model is not perfect. The platform's own transparency report cited in April 2026 disclosed a false-positive rate around 4 to 6 percent across the five categories combined, with the highest false-positive rate in the medical content category where clinical imagery is often classified alongside gore. Creators in regulated categories — healthcare professionals, harm reduction educators, forensic researchers — face higher friction and are encouraged to apply a manual sensitive flag before publication so the platform can route the content through the educational-context exception. For advertiser-side ad rejection prediction tooling, see Meta Rejection Predictor .

What's the difference between X's sensitive media flag and the ACC adult content program?

The sensitive media flag and the Adult Content Creator program govern different content layers with different eligibility requirements and different commercial consequences. The sensitive media flag is a platform-wide content classification that applies to any user posting any of the five sensitive media families and operates through a single visibility control — content is either shown or hidden depending on the viewer's account setting and the platform's defaults. There is no monetisation pathway attached to the sensitive media flag in itself, and no eligibility threshold to access it. Any account can post sensitive media, subject to the policy boundaries, and any account can choose to hide it from view. The Adult Content Creator program is a parallel framework that applies specifically to nudity and sexual content above the suggestive threshold and operates as an opt-in creator program with eligibility requirements including age verification, account history, and creator agreement to a separate set of content rules. Creators in the ACC program are subject to three-tier content classification — non-explicit, explicit, and explicit-extreme — with different visibility defaults and different monetisation eligibility at each tier. The ACC program also operates the Adjacency Shield mechanism for advertisers, which provides a stronger separation between adult content and advertiser inventory than the standard sensitive media flag does. The practical implication is that a creator publishing suggestive but non-adult content sits under the sensitive media flag rather than ACC, while a creator publishing nudity or sexual content must be in ACC to publish lawfully. For the dedicated guide to the adult content side, see X Adult Content Policy 2026 Rules Guide .

How can advertisers control ad adjacency to sensitive media in 2026?

X provides three distinct advertiser controls for sensitive media adjacency in 2026, each with different granularity and different brand safety profiles. The first control is the platform-wide sensitive media adjacency opt-out, which is on by default for all advertiser accounts and prevents standard ad inventory from appearing adjacent to content flagged under any of the five sensitive media families. The opt-out applies at the campaign level and can be reviewed in the advertiser dashboard under brand safety controls. The second control is the category-level granular opt-in, which is available to advertisers in specific categories where a degree of sensitive content adjacency is acceptable or even desirable — harm reduction, public health awareness, regulated industries that need to advertise alongside relevant medical content, news and current affairs advertisers that need to appear alongside conflict reporting. Granular opt-in is not available for the adult content tier and is subject to creative review before activation. The third control is the third-party brand safety integration, where X exposes its sensitive media classification through its brand safety partners — DoubleVerify, Integral Ad Science, and the platform's own native brand safety reports. The integration provides post-campaign reporting on adjacency events and supports adjacency exclusion lists at the publisher account level for advertisers that want granular control over which specific accounts their ads can appear alongside. Advertisers operating in regulated industries should treat the default opt-out as the floor rather than the ceiling and should configure category-level granular controls where their compliance posture requires tighter separation. For the consolidated advertiser policy reference, see X Ads Policy .

What EU DSA requirements apply to X's sensitive media settings for minors?

The EU Digital Services Act produces three specific requirements that govern how X handles sensitive media for minors and how the platform configures defaults for EU accounts. The first requirement is that minors cannot be subject to advertising based on profiling under Article 28, which has produced platform-wide changes including X's removal of behavioural ad targeting for accounts the platform has confirmed or strongly inferred to be under 18. The behavioural ad targeting removal applies regardless of the sensitive media setting and is separate from the sensitive media adjacency controls that advertisers configure. The second requirement is that the recommender system must not amplify content that poses risks to minors' physical or mental wellbeing, which X has implemented in part through a default 'hide sensitive media' setting for any account whose age signal places it below 18. The default cannot be toggled off on minor-confirmed accounts, and the platform's confirmation rules combine the age stated at signup with secondary signals from account behaviour. The third requirement is the DSA risk assessment obligation under Article 34, which requires X to assess the systemic risks that sensitive media surfaces produce for minors and to document the mitigation measures in place. The platform's 2026 risk assessment summary, published under the Article 9 transparency obligation, identifies the auto-detection coverage gaps in the medical and suggestive categories as the residual risk areas that require ongoing model improvement. Advertisers should expect that the EU regulatory floor on minor protection will tighten over the course of 2026 as the European Board for Digital Services issues guidance that interprets the recommender system obligations in light of accumulated enforcement data. For consolidated DSA framework reference, see European Union DSA Compliance .

How long does the appeal process take and what's the reversal rate?

X's sensitive media appeal process operates on a tiered timeline that depends on the appeal type, the account history, and whether the appeal involves manual review or qualifies for the priority review queue. Standard appeals — submitted by an account with no prior policy violations and concerning a single piece of flagged content — typically resolve within 24 to 72 hours under the 2026 service levels. Expanded appeals — concerning multiple pieces of content, account-level visibility restrictions, or repeat flag patterns — typically resolve within 5 to 10 business days. Edge case appeals — concerning the medical content exception, the educational context exception, or the news and current affairs exception — typically require manual specialist review and can take up to 21 business days. The platform's transparency report cited in April 2026 disclosed a reversal rate around 22 percent for standard appeals and around 35 percent for medical content category appeals where the educational context exception is properly invoked. The reversal rates reflect the false-positive friction in the auto-detection model and the platform's willingness to recognise context arguments that the initial classifier missed. Creators who anticipate friction in regulated content categories can pre-empt some of the appeal cycle by applying a manual sensitive flag at publication time and including a clear educational, journalistic, or harm-reduction context statement in the post text. The pre-emptive approach does not eliminate the auto-detection pass, but it does reduce the rate at which content is restricted before appeal. For automated content language audit support, see Keyword Risk Checker .

X Sensitive Media Settings 2026: Rules, Flags & Brand Safety

What X's 2026 Sensitive Media Policy Covers

X's 2026 sensitive media framework is the platform's general-purpose content classification layer for material that does not meet the adult content threshold but exceeds what the platform considers safe for default display. The framework sits alongside — and is operationally distinct from — the Adult Content Creator program, which governs nudity and sexual content under a separate set of eligibility rules and tier classifications. Conflating the two leads to compliance errors on both sides, which is why this guide treats them as separate workflows.

The sensitive media policy applies to any account posting content that falls inside one of five category families. The same flag drives the same downstream behaviours regardless of which family triggered it — the content is hidden by default for viewers with sensitive media off, advertiser inventory is excluded from adjacency, and a small set of recommender system constraints apply. The simplicity of the downstream rules belies the complexity of the classification, which depends on a hybrid auto-detection model that has changed significantly through 2025 and 2026.

"Sensitive media is the layer where the largest share of policy disputes resolve, because the line between sensitive and not-sensitive is more contested than the line between adult content and not-adult-content."
— X Trust & Safety, April 2026 transparency report

For advertisers, the practical compliance question is not whether sensitive media exists on the platform — it does, in volume — but whether their ad inventory can end up adjacent to it, and what controls they need to configure to bring the adjacency rate down to a defensible level for their brand safety policy. For creators, the practical question is whether their content will be flagged, what their options are if it is, and how to navigate the appeal process when the classification is wrong.

Sensitive Media Categories: What Triggers a Flag

The five category families that X classifies as sensitive media each have distinct trigger criteria, distinct false-positive profiles, and distinct downstream consequences for advertiser adjacency. Understanding the categories matters because the appeal process and the educational-context exception apply differently to each.

Category	Primary Trigger	Common False Positive	Educational Exception?
Graphic Violence	Injury, death, weapons in active use, aftermath of violent events	Historical or news imagery without explicit gore	News, journalism, conflict documentation
Hyper-Realistic Gore	Wound photography, forensic imagery, surgical content out of context	Medical education imagery, harm reduction visuals	Medical, public health, harm reduction
Suggestive Content	Partial nudity in artistic context, intimate imagery without explicit content	Fitness content, fashion editorial, wellness imagery	Limited — narrowly drawn
Medical Content	Clinical procedures, severe pathology, emergency medicine documentation	Healthcare professional education	Yes — verified healthcare context
Legacy Adult (now under ACC)	Nudity, sexual content above suggestive threshold	Misclassified suggestive content escalated incorrectly	None — moved to ACC framework

The legacy adult category is operationally separate from the other four. Content that meets the adult threshold must be published from an Adult Content Creator account and is subject to the ACC tier classification, not the standard sensitive media flag. The cross-traffic between the categories happens primarily at the suggestive-to-adult boundary, where the classifier sometimes escalates suggestive content into the adult bucket — a misclassification that the appeals process is configured to correct.

Joint Text-Image Classification

The 2026 model treats text and image jointly. A clinical image accompanied by clinical text typically clears the medical category. The same image accompanied by sensationalising or suggestive text gets escalated. The reverse direction also operates — a benign image accompanied by suggestive text can be classified as suggestive content even when the image alone would not trigger a flag. Creators in regulated categories should treat the post text as part of the classification surface, not as a separate channel.

Auto-Detection vs Manual Flag

X operates a two-track classification pipeline. The auto-detection track runs on every upload and applies the model classification before publication. The manual flag track is creator-controlled and allows a creator to apply a sensitive flag voluntarily, which routes the content through the educational-context exception logic and reduces the rate at which the post is subsequently restricted by auto-detection.

Auto-Detection Pipeline

Stage one is content recognition. The platform's image and video classification model evaluates the visual content against the five sensitive media families and outputs a per-category confidence score. The model is retrained on a quarterly cycle and incorporates feedback from the appeal process and from the platform's trust and safety reviewers.

Stage two is context fusion. The model takes the post text, the embedded link target if any, the conversation thread, and the account's prior behaviour into account, adjusting the per-category score. A creator with a clean history posting in a clearly educational context will see lower friction than a new account posting similar content.

Stage three is policy application. The final classification is mapped to the sensitive media flag, the default visibility rule, and the advertiser adjacency status. The post is either published with the flag attached or held for manual review if the confidence is in a contested band.

Manual Flag Pre-Emption

Creators can apply a sensitive flag at publication time through the compose interface. The manual flag does not bypass the auto-detection pass, but it does change two things. First, the platform interprets the manual flag as a context signal that informs the educational-context exception logic for medical, harm reduction, journalistic, and conflict documentation content. Second, the rate of post-publication retroactive reclassification is materially lower for manually-flagged content because the creator has already accepted the classification and the model treats the post as settled.

False-Positive Profile

The April 2026 transparency report disclosed an aggregate false-positive rate around 4 to 6 percent across the five families combined. The medical content category has the highest false-positive rate, primarily because clinical imagery is visually similar to gore and the educational-context exception requires manual specialist review to invoke. Healthcare professionals, harm reduction educators, and forensic researchers face the highest friction and benefit most from pre-emptive manual flag application.

User Settings: Sensitive Media Display Controls

The viewer-side controls have shifted in 2026. The default state for new accounts now depends on jurisdiction and confirmed age, and the override controls have narrowed for accounts the platform has confirmed or inferred to be under 18.

Default State by Jurisdiction

EU and UK accounts: 'Hide sensitive media' is the default for all new accounts. The default reflects the DSA recommender system obligation under Article 34, which requires platforms to mitigate systemic risks to minors and adult users.
US and other accounts: 'Display sensitive media with warning' remains the default, with a one-tap reveal mechanism for individual posts.
Minor-confirmed accounts (all jurisdictions): 'Hide sensitive media' is mandatory and cannot be toggled off. The minor-confirmed status combines the age stated at signup with secondary inference signals from account behaviour.

Override and Reveal Mechanisms

Adult users who have toggled sensitive media display on still see the warning overlay on individual posts and must tap to reveal. The reveal action is logged and counts toward the platform's user-level engagement signals for the recommender system. Users can also configure category-level visibility — for example, choosing to display medical content but hide graphic violence — through the granular sensitive media controls in account settings.

Advertiser Brand Safety & Ad Adjacency

Advertiser controls for sensitive media adjacency operate at three layers — the platform-wide default, the category-level granular configuration, and the third-party brand safety integration. Each layer addresses a different brand safety question and produces different reporting outputs.

Platform-Wide Default Opt-Out

The platform-wide opt-out is on by default for every advertiser account and prevents standard ad inventory from appearing adjacent to any content flagged under any of the five sensitive media families. The opt-out applies at the campaign level and is visible in the advertiser dashboard under brand safety controls. Advertisers running in standard inventory should treat the default opt-out as the compliance floor.

Category-Level Granular Opt-In

Advertisers in specific categories may want to allow adjacency to certain sensitive media families — harm reduction nonprofits advertising alongside harm reduction educational content, news advertisers appearing alongside conflict documentation, regulated medical advertisers appearing alongside clinical content. The category-level granular opt-in supports this with per-category toggles, subject to a creative review pass before activation. The granular opt-in is not available for the legacy adult category, which is governed by the ACC Adjacency Shield separately.

Third-Party Brand Safety Integration

X exposes its sensitive media classification through DoubleVerify, Integral Ad Science, and the platform's native brand safety partners. The integration provides post-campaign reporting on adjacency events and supports publisher account exclusion lists for advertisers that want granular control over which specific accounts their inventory can appear alongside. Brand safety reporting includes per-category adjacency rates and supports audit-trail evidence for advertiser internal compliance reviews.

Industry-Specific Considerations

Healthcare advertisers should configure granular opt-in for medical content adjacency where the campaign objective supports it, while excluding suggestive content adjacency entirely. Financial services advertisers should exclude all sensitive categories and rely on the default opt-out plus an exclusion list. News and current affairs advertisers should configure granular opt-in for graphic violence in journalistic context while excluding suggestive content. Children's product advertisers should treat the default opt-out as the floor and additionally apply minor-protection controls through the platform's audience configuration. For consolidated industry context, see Healthcare Social Media Compliance.

EU DSA Defaults & Minor Protection

The EU Digital Services Act has reshaped the default user experience for EU and UK accounts and has produced specific obligations for X under Article 28 (advertising to minors), Article 34 (systemic risk assessment), and Article 35 (mitigation measures). The 2026 sensitive media defaults reflect those obligations and continue to evolve as the European Board for Digital Services issues interpretive guidance.

Article 28 — Advertising to Minors

Behavioural advertising based on profiling cannot be served to accounts the platform has confirmed or strongly inferred to be under 18. The prohibition applies regardless of the sensitive media setting and is operationalised through the platform's audience configuration. Advertisers cannot opt out of this rule, and platform-level enforcement removes minor-inferred accounts from behavioural audiences automatically.

Article 34 — Systemic Risk Assessment

X is required to assess the systemic risks that sensitive media surfaces produce for users in the EU and to document the mitigation measures in place. The 2026 risk assessment summary, published under the Article 9 transparency obligation, identifies the auto-detection coverage gaps in the medical and suggestive categories as the residual risk areas requiring ongoing model improvement.

Article 35 — Mitigation Measures

Mitigation measures that the platform has implemented include the default 'hide sensitive media' setting for new EU accounts, the mandatory hide setting for minor-confirmed accounts, the recommender system constraints on sensitive media amplification, and the brand safety reporting infrastructure that supports advertiser oversight.

Compliance Posture for EU-Operating Brands

EU-operating brands should treat the DSA framework as the regulatory floor rather than the ceiling. Specifically: configure the default opt-out as the campaign baseline, apply category-level granular controls where the campaign profile requires sensitive content adjacency, exclude minor-inferred audiences from any campaign that touches sensitive media adjacency even in granular opt-in, document the brand safety configuration in the internal compliance record, and review the configuration quarterly against the platform's updated transparency reports.

Appeals Process & Reversal Timelines

X's appeal process operates on a tiered timeline that varies by appeal type, account history, and category. The 2026 service levels are tighter than the 2024 baseline and the reversal rates have improved as the auto-detection model has been retrained on appeal-derived data.

Appeal Type	Typical Timeline	Reversal Rate (Q1 2026)
Standard (single post, clean account)	24 to 72 hours	~22%
Expanded (multiple posts or account-level restriction)	5 to 10 business days	~18%
Edge case (medical / educational context)	Up to 21 business days	~35%
Advertiser brand safety dispute	5 business days	~12%

Reducing Appeal Friction

Pre-emptive manual flag: Apply the sensitive flag at publication time for any content that sits in a contested category. The flag does not bypass auto-detection but it does reduce retroactive reclassification.
Context statement: Include a brief educational, journalistic, or harm-reduction context statement in the post text. The context fusion stage of the model reads the post text as a classification signal.
Account history hygiene: Clean account history reduces the friction multiplier on every classification decision. Accounts with prior violations face stricter thresholds and longer appeal timelines.
Verified context: For medical and harm reduction creators, verified professional context through account verification or organisational affiliation invokes the educational-context exception more reliably.

Compliance Checklist

[ ] Reviewed which of the five sensitive media families could apply to your content or campaigns
[ ] Configured advertiser default opt-out as the campaign baseline
[ ] Applied category-level granular opt-in only where campaign profile requires it
[ ] Excluded minor-inferred audiences from any campaign touching sensitive media adjacency
[ ] Configured third-party brand safety integration for adjacency reporting
[ ] Documented brand safety configuration in internal compliance record
[ ] For EU operation: aligned defaults with DSA Article 34 systemic risk framework
[ ] Creator-side: trained content team on pre-emptive manual flag for contested categories
[ ] Creator-side: added educational, journalistic, or harm-reduction context statements where applicable
[ ] Healthcare creators: secured verified professional context for educational exception
[ ] Established appeal escalation process for false-positive flag decisions
[ ] Reviewed quarterly transparency reports for adjacency rate trends
[ ] Cross-referenced this workflow with adult content creator program where ACC eligibility applies
[ ] Updated record of processing activities for the sensitive media classification surface

For complementary creator-side compliance reference, see X Adult Content Policy 2026 Rules Guide and for ongoing policy change tracking, see Policy Tracker.

Don't miss the next policy change.

Subscribe to the Policy Tracker — get weekly digests or instant Pro alerts across all 8 platforms. Or try our free Keyword Risk Checker first.

Subscribe Free

Report Keywords — Run AI Compliance Audit

#X Ads Policy#Sensitive Media#Brand Safety#Content Moderation#Ad Adjacency#DSA#Age Gating#Image Recognition#Advertisers#Creators#Compliance Guide 2026#X Twitter

Share This Report

Tweet Share

X Age Verification Requirements 2026: Process, Issues & Appeals Guide

X's 2026 age verification has expanded across adult content, sensitive media, EU recommender systems, and advertiser audience configuration. Here's the full process.

X Community Notes on Paid Ads Enforcement 2026 — Advertiser Response Framework, Context Labels & Brand Safety Implications

X extended Community Notes to paid ads on April 19, 2026 allowing contributors to add context labels to sponsored posts, reshaping advertiser response workflows and brand safety expectations.

X (Twitter) Hateful Conduct & Content Moderation Policy 2026 — Rules, Enforcement Changes & Advertiser Impact

X's hateful conduct policy has undergone significant changes in 2026. This compliance guide covers current rules, enforcement tiers, advertiser brand safety tools, and actionable steps to protect campaigns on the platform.

X Sensitive Media Settings 2026: Auto-Detection, Flags & Brand Safety