Log entry Nº 208 · the stack watch

Bearing N 000° · November 11, 2026 · 3 min read

Advertising TechN 000°Watch log

X Previews AI Labels Explaining Flags and Appeal Paths

X is previewing AI-generated labels that reveal why a post was flagged, how its reach changed, and how users can appeal the decision at a New York safety summit.

By Rebecca Stone3 min read537 words

X Previews Labels That Tell Users Why Its AI System Flagged Their Posts
Fix Nº 208X Previews Labels That Tell Users Why Its AI System Flagged Their Posts — AI-generated

By Rebecca Stone

On Nov. 10, X previewed AI-generated labels with three disclosure elements.

A post once carried the label “adult content.” It featured an image of a SpaceX rocket launch. No one, including the author, could see why.

That opaque outcome became the problem Michael O’Herlihy wants to fix. O’Herlihy is a Twitter veteran who recently rejoined X as head of safety.

At the Brand & AI Safety Summit in New York on Tuesday, Nov. 10, he presented early designs. The proposed labels would appear beneath content on X. He said the systems may show their decisioning process, rather than leaving users with an unexplained flag.

The designs cover three practical details:

  • What X’s AI detected in the post.
  • How the post’s reach was limited after it was flagged.
  • How the user can appeal the decision.

O’Herlihy said the designs would “explain what its AI detected in a post, detail how exactly the post’s reach was limited after being flagged, and allow users to appeal the decision.” That combines a visible explanation with a route to challenge it.

What does this change for brand-safety teams?

For content and advertising teams, the change is visibility. It joins three points that usually remain separate: the detected issue, the distribution consequence, and available recourse.

A moderator or brand-safety reviewer can see what triggered the flag. They can also see how reach was limited. A creator receives a stated appeal route. The preview does not guarantee a favorable appeal, but it makes the decision reviewable.

For a brand-safety team, the appeal mechanism matters as much as the rationale because it turns a passive disclosure into a documented challenge that X must handle.

O’Herlihy’s broader argument supports that model. He argues that AI-led moderation creates auditable records better than human review. The SpaceX example shows the practical value of such a record. An explanation can expose a mistaken detection. Reach details can show the commercial effect of the flag.

That value also carries a sales message. X’s advertising revenue has plummeted since Elon Musk took over. Winning brands back makes brand safety a crucial part of the pitch.

Labels documenting a detection, its effect on distribution, and an appeal route give X a more concrete brand-safety story. They show what the company means by moderation accountability, although the designs remain a preview.

The announcement does not set a general rollout date. It does not say when every user will see the labels. Nor does it specify supported languages, appeal eligibility, response times, or whether an appeal can reverse a flag. It provides no figures for advertising revenue, label accuracy, appeal outcomes, or post reach after moderation.

Those missing details limit what practitioners can conclude today. The announcement establishes an intended structure, not a complete service commitment. X has shown the direction: rationale, reach limitation, and appeal mechanism beneath a flagged post.

For brands and agencies, the immediate task is to watch what becomes available. They should test whether each disclosure works in practice. X’s proposal makes moderation more inspectable. Its commercial claim will depend on whether those records give advertisers and creators evidence they can use.

Plot points

  1. On Nov. 10, Michael O’Herlihy previewed AI-generated moderation labels at the Brand & AI Safety Summit in New York.
  2. Three proposed disclosure elements cover what the AI detected, how reach was limited, and how users can appeal.
  3. A SpaceX rocket-launch image was once labeled “adult content” without a visible explanation.
  4. O’Herlihy argues that AI-led moderation creates auditable records better than human review.
  5. X’s advertising revenue has plummeted since Elon Musk took over; brand safety is central to winning brands back.

via adweek.com (Source)

Filed under

  • x
  • brand-safety
  • ai-moderation
  • content-labeling
  • advertising-transparency
Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Staff writer covering companies, deals and strategy in marketing technology at Martech Beacon.

38 articles

Neighbouring fixes

‹ Previous articleNext article ›