Elon Musk teases a new image-labeling system for X… we think?

Elon Musk appears to be hinting at a new X feature that would label edited images as “manipulated media,” though details on how the system works remain unclear.

So far, the only indication of the feature comes from a brief and cryptic post by Musk himself. In the post, Musk wrote “Edited visuals warning” while resharing an announcement from the anonymous X account DogeDesigner. That account is frequently used as an unofficial channel for rolling out or previewing new X features, with Musk often amplifying its posts.

Beyond that, specifics are scarce. DogeDesigner claimed the feature would make it “harder for legacy media groups to spread misleading clips or pictures,” and described it as a new addition to the platform. However, neither Musk nor X has explained how images will be classified as edited, manipulated, or AI-generated — or whether the system applies only to generative AI imagery or also to content altered using traditional tools such as Adobe Photoshop.

Before Musk acquired the platform and renamed it X, Twitter had a policy for labelling manipulated or deceptively altered media. Rather than removing such posts outright, Twitter applied warning labels to tweets containing misleading visuals. That policy was not limited to AI-generated content. In 2020, former site integrity head Yoel Roth said the rules also covered practices such as selective cropping, slowing down video, overdubbing audio, or altering subtitles.

It is unclear whether X is reviving that framework, modifying it, or introducing an entirely new system focused on AI. Current help documentation on X references a policy against sharing inauthentic media, but enforcement has been inconsistent. That gap was highlighted recently when non-consensual deepfake nude images circulated widely on the platform with little apparent intervention. Even the White House has shared manipulated images in recent months, further blurring the lines.

Edited visuals warning https://t.co/0OIz5PvwSz — Elon Musk (@elonmusk) January 28, 2026

Labelling content as “manipulated media” or “AI-generated” is rarely straightforward. Given X’s role as a central hub for political messaging and propaganda — both domestic and international — clarity around how the platform defines and detects edited visuals is critical. Users would also need to know whether there is any appeals process beyond the platform’s crowdsourced Community Notes system.

Other platforms have already learned how difficult this can be. When Meta rolled out AI image labels in 2024, its detection systems frequently misfired. In several cases, Meta incorrectly labelled genuine photographs as “Made with AI,” even though they were not generated with generative tools.

Those errors were primarily caused by the increasing integration of AI-powered features into standard creative software. Tools commonly used by photographers and designers now incorporate AI in subtle ways, which can confuse automated detection systems. One example involved Adobe’s cropping tool, which flattened images before saving them as JPEGs — a process that inadvertently triggered Meta’s AI detection. In other cases, Adobe’s Generative Fill feature, used to remove minor visual imperfections, caused images to be labelled as AI-generated even when the original photo remained largely intact.

In response, Meta revised its labelling language to a more neutral “AI info” tag, rather than asserting that AI definitively created images.

There are also broader industry efforts to address this challenge. The C2PA (Coalition for Content Provenance and Authenticity) works on standards for verifying the origin and editing history of digital media. Related initiatives include the Content Authenticity Initiative and Project Origin, both of which focus on embedding tamper-evident provenance metadata into images and video.

Major technology companies — including Microsoft, the BBC, Adobe, Arm, Intel, Sony, OpenAI, and others — sit on C2PA’s steering committee, with many more participating as members. Platforms such as TikTok have also begun labelling AI-generated content, while streaming services like Deezer and Spotify are scaling systems to identify and flag AI-generated music. Google Photos has adopted C2PA standards to show how images were created or edited.

X, however, is not currently listed as a member of C2PA. It remains unclear whether the company has joined the initiative recently or plans to rely on an alternative approach. Musk did not specify whether the teased feature is designed to detect AI-generated images specifically or applies more broadly to any visual content that has been altered before being uploaded. It is also uncertain whether the feature is genuinely new, despite DogeDesigner’s claims.

X typically does not respond to media inquiries. Still, questions remain about how — or even whether — the platform plans to formalise its approach to labelling manipulated media, as the line between traditional editing and AI-assisted creation continues to blur.