Musk’s Grok chatbot restricts image generation after global backlash to sexualized deepfakes

Workers install lighting on an "X" sign atop the company headquarters, formerly known as Twitter, in downtown San Francisco, July 28, 2023. (AP Photo/Noah Berger)

Elon Musk’s AI chatbot Grok is preventing non-paying users from generating or editing images after a global backlash erupted over sexualized deepfakes of people, but the change has not satisfied authorities in Europe.

The chatbot, which is accessed through Musk’s social media platform X, has in the past few weeks been granting a wave of what researchers say are malicious user requests to modify images, including putting women in bikinis or in sexually explicit positions.

Researchers have warned that in a few cases, some images appeared to depict children. Governments around the world have condemned the platform and opened investigations.

On Friday, Grok responded to image altering requests with the message: “Image generation and editing are currently limited to paying subscribers. You can subscribe to unlock these features.”

While subscriber numbers for Grok aren’t publicly available, there was a noticeable decline Friday in the number of explicit deepfakes that Grok is now generating compared with just days earlier.

Grok was still granting image requests but only from X users with blue checkmarks given to premium subscriber who pay US$8 a month for features including higher usage limits for the chatbot.

The Associated Press on Friday confirmed Friday afternoon that the image editing tool was still available to free users on the standalone Grok website and app.

An X spokesperson didn’t respond immediately to a request for comment.

The restrictions for users save for paying subscribers did not appear to change the opinions of leaders or regulators in Europe.

“This doesn’t change our fundamental issue. Paid subscription or non-paid subscription, we don’t want to see such images. It’s as simple as that,” said Thomas Regnier, a spokesman for the European Union’s executive Commission. The Commission had earlier slammed Grok for “illegal” and “appalling” behaviour.

The British government was also unsatisfied.

Grok’s changes are “not a solution,” said Geraint Ellis, a spokesman for British Prime Minister Keir Starmer, who on Thursday had threatened unspecified action against X.

“In fact, it is insulting to the victims of misogyny and sexual violence,” he said, noting that it shows that X “can move swiftly when it wants to do so.”

“We expect rapid action,” he said, adding that “all options are on the table.”

Starmer, speaking to Greatest Hits radio, had said that X needs to “get their act together and get this material down. We will take action on this because it’s simply not tolerable.”

The U.K.’s media and privacy regulators both said this week they’ve contacted X and Musk’s artificial intelligence company xAI for information on measures taken to comply with British regulations.

France, Malaysia and India have also been scrutinizing the platform and a Brazilian lawmaker has called for an investigation. The European Commission has ordered X to retain all internal documents and data relating to Grok until the end of 2026, as part of a wider investigation under the EU’s digital safety law.

Grok is free to use for X users, who can ask it questions on the social media platform. They can either tag it in posts they’ve directly created or in replies to posts from other users.

Grok launched in 2023. Last summer the company added an image generator feature, Grok Imagine, that included a so-called “spicy mode” that can generate adult content.

The problem is amplified both because Musk pitches his chatbot as an edgier alternative to rivals with more safeguards, and because Grok’s images are publicly visible, and can therefore be easily spread.

Kelvin Chan, The Associated Press