X tests detailed AI content labels: users should see the reason

X is developing more detailed labels for AI-generated posts. The labels should not only indicate the use of artificial intelligence, but also explain why a post was classified accordingly. Michael O’Herlihy, who is responsible for Brand Safety at X, presented the first examples on 10 November 2026 at the Brand & AI Safety Summit in New York.

The difference lies in the reasoning

According to a report by Social Media Today, X is working on automated detection that is meant to distinguish more clearly between forms of AI use. Users are to receive hints about what led to the classification. For creators this is especially relevant when they believe a post was wrongly flagged: a justified label is easier to contest than an undifferentiated warning symbol.

The labels shown so far are test examples from a presentation, not a confirmed global rollout. No binding start date has been named.

The topic differs from the already announced X notices on restricted posts and reach: those concern moderation measures, whereas this concerns specifically the origin and AI editing of content.

Marketing needs more than an AI symbol

There is a world of difference between a fully synthetic recording and a real video whose subtitles were translated by AI. For brands, more precise labelling would help, because it makes technical editing and potentially misleading presentation easier to tell apart. X already supports Grok-powered subtitle correction and translation for videos – an example of AI in everyday production workflows.

A good label should answer two questions: What was done with AI? And why is this information important for assessing the post? This applies especially to news images, political motifs and testimonials. Around the US elections it already became clear how synthetic advertising materials make labelling harder.

European transparency rules increase the pressure

For German companies, the development is not solely a platform question. The European AI Act sets out transparency obligations for certain artificially generated and manipulated content. The Bundesnetzagentur explains the rules and the associated transparency code.

The practical consequence for social teams: origin, editing steps and consents should be documented already during production. Anyone who distributes the same clip across several platforms can thus explain what was actually recorded and what was synthetically altered. Other platforms are now also relying on AI labels for music.

X is trying to attach this explanation directly to the post. Whether the detection works reliably and whether misattributions can be corrected easily will have to be shown in practice.