Scandal over sexualized deepfakes: Grok largely refuses image generation
On X, Grok began rejecting requests for image generation on Friday morning. A paid subscription is now required for this, according to the responses.
(Image: miss.cabul/Shutterstock.com)
Grok on X is currently no longer generating images for users who do not pay for the microblogging service and therefore can no longer be used to digitally “undress” people. This is evident from responses from the AI chatbot, which it has been generating since Friday morning in response to requests for image editing. It states, “Image generation and editing is currently limited to paying customers.” There is also a link to the Premium subscription for X, which costs 9.44 euros per month. There is no further statement or information on this step. X CEO Elon Musk has also not commented on it.
The move by X is likely a reaction to the outrage over the handling of Grok, which has increased massively in recent days. For days, users have been using the AI account to digitally “undress” thousands of photos, primarily of women, per hour. This means the AI created versions of images of clothed individuals for them, on which they appeared to be wearing a bikini or underwear. Minors were also affected. For those affected, who in most cases did not give their consent, this can have extremely severe consequences. Such technology is prohibited in some states. However, this happened in full public view on X.
No details from X or xAI
What exactly the extensive deactivation of the function now means is unclear. Because it has not been completely removed, paying users can still order such content. However, they are identifiable to X and can be subject to consequences. Shortly before the shutdown, the US magazine Wired pointed out that a loophole on Grok's website makes significantly more disturbing content accessible, which users can have created there. In viewable content from there, photorealistic AI-generated video clips with sexualized deep fakes, to which blood is also added, can be found. Clips that appear to show prominent women are also said to be included. There are even short films that appear to show minors.
Videos by heise
Just on Thursday, CNN also reported that Elon Musk had internally railed against limitations for Grok for a long time, which were intended to prevent the chatbot from creating problematic content. The billionaire was reportedly dissatisfied with “excessive censorship,” the US news channel reported, citing several sources. In a meeting, he reportedly showed particular dissatisfaction and criticized the restrictions. Around this time, three senior employees of the AI company xAI, responsible for Grok, resigned. Musk had repeatedly promoted on X that Grok should also comply with more explicit requests in “Spicy Mode.”
(mho)
