X introduces Grok's Edit Image Function, Risks Creative Properties

Screenshot of Elon Musk's announcement of the Edit Image function on Decemeber 25 / X

On Christmas Day, the social media platform X (formerly Twitter) launched an expansion of its Grok AI capabilities, introducing a feature that allows users to edit any image posted on the platform using generative text prompts.

Elon Musk, owner of X, described the tool as a way for users to "imagine new content" and "correct imperfections." However, the rollout has immediately sparked a firestorm among digital artists, who argue the tool leads to unauthorized modifications of their work and provides a streamlined method for removing watermarks.

How It Works

The "Edit Image" function is currently available on the desktop version of X.

Users can access it by clicking on an image's properties or selecting the "Edit Image" button directly on a post. Notably, the tool remains functional even if the original content creator has opted out of allowing Grok to use their account data for AI training.

Tests conducted by Kuhaku Media confirmed how it manipulates images, noting that the function remains available even for posts containing multiple images.

Comparison between an image sample of Kuhaku Media's original coverage (Left) and a prompted sample with Grok's Edit Image function with a prompt that removes Kuhaku's watermark and forces the people to look at the camera (Right).

Poison Immunity

X users who use platforms like Glaze and Nightshade, which artists use to protect their digital intellectual property from AI scraping through Poisoning AI don't work against Grok's Edit Image function.

Screenshot from @SpookyZombae / X

Screenshot from @VAMichaelaLaws / X

Image poisoning through the use of Glaze and Nightshade distorts the image submitted on the platform to make it unusuable by any Artificial Intelligence while still being indistinguishable by humans.

Illustration of Image Poisoning by comparing original images as clean models and poisoned images / Cornell University.

How the Art Community Responds

In response to the feature being able to bypass typical poisoning of both Nightshade and Glaze, artists are scrambling to safeguard their portfolios by deleting their previous works, turning their posts into GIFs or static videos to prevent the edit image function being available on the post, or using links on images.

X users @O2H2_OH4 and Vgen's co-founder @_muncha_ using GIF's and clickable links to disable Grok's Edit Image function

Several artists have also reported that they have been using websites that allow them to crosspost from Bluesky where an image posted to X would lead them to a Blusky link.

X user @alt_fogado encouraging users to use a platform that enables crossposting from social media application, Bluesky.

While text-to-image manipulation has existed since 2020, X is one of the major social media platform to integrate such powerful generative editing tools directly into its user interface.

Artists warn that the tool increases instances for image theft, the creation of non-consensual pornographic content, and the spread of visual misinformation all within a platform ecosystem that previously relied on the integrity of uploaded media.