X teases new labeling system for manipulated media

X is reportedly rolling out a feature to label edited images as manipulated media, if its owner, Elon Musk, is to be believed. The rollout remains unconfirmed by official company announcements, relying instead on a cryptic post from Musk. Resharing a post from DogeDesigner—an anonymous X account used as a proxy for introducing new features—Musk added the comment “Edited visuals warning”. According to DogeDesigner, the feature is new to X and could make it “harder for legacy media groups to spread misleading clips or pictures.” However, X has not clarified how it will determine what constitutes edited or AI-generated content, or whether the feature applies to images edited with traditional tools.

The platform has a historical precedent for content moderation, though its enforcement has been inconsistent. In 2020, Yoel Roth, the former site integrity head at Twitter (before the platform was acquired and renamed as X), commented on the company’s policy regarding manipulated media, which labeled deceptively altered or fabricated media as an alternative to removal. Currently, X’s help documentation says there’s a policy against sharing inauthentic media. However, this policy is rarely enforced, as highlighted by the recent deepfake debacle of users sharing non-consensual nude images.

The lack of clarity on how X defines edited or AI-generated content raises significant questions about accuracy and potential enforcement. This is particularly critical given that X operates as a playground for political propaganda, both domestically and abroad. Without documented guidelines or a clear dispute process beyond X’s crowdsourced Community Notes, users are left with little understanding of how the platform will identify manipulated media.

Other platforms are already moving forward with standardized labeling. In 2024, Meta introduced AI image labeling, while TikTok has also been labeling AI content. Streaming services like Deezer and Spotify are scaling initiatives to label AI music. Many of these platforms rely on the Coalition for Content Provenance and Authenticity (C2PA), a standards-setting body for verifying content provenance. Google Photos uses C2PA to indicate photo provenance, and the steering committee of the C2PA includes Microsoft, the BBC, Adobe, Arm, Intel, Sony, and OpenAI. X, however, is not currently listed as a member of the coalition, and the company has not clarified its technical approach to identifying edited media.

Why it matters

X is introducing a ‘manipulated media’ labeling feature, though the company has not clarified how it will determine what constitutes edited or AI-generated content, raising questions about potential enforcement and accuracy.