Musk's Grok AI Lacks Basic Guardrails for Sexual Deepfakes, Allows Users to Generate Porn-Like Videos

Elon Musk’s AI chatbot, Grok, is facing major criticism after a new “spicy” setting on its image generator was found to create explicit and inappropriate deepfake images of women. The new Grok feature, called "Imagine," allows users to create AI-generated video clips with a "spicy mode" that can generate nudity. 

This tool has been used to create and share harmful deepfake content, which are digitally altered videos and images of people without their consent. The issue has especially been brought to the spotlight because of its use to create inappropriate content around celebrities like Taylor Swift. 

The Taylor Swift Deepfake Incidents

According to The Verge, when Grok was prompted to create a video of "Taylor Swift celebrating Coachella," the AI tool went beyond the request. It generated a series of revealing images and a deepfake video of her dancing topless without being prompted for nudity. The Verge’s reporter Jess Weatherbed said Grok Imagine “didn’t hesitate to spit out fully uncensored topless videos."

Especially concerning is Grok’s sexualization of almost exclusively women. As Matt Novak from the Gizmado wrote after running several prompts, “Videos of men were the kind of thing that wouldn’t really raise many eyebrows.” 

Rather it was only women that Grok would engage in creating inappropriate images, reflecting the statistic that 90% of image based sexual abuse victims are women. Grok created such images for many female celebrities beyond Taylor Swift. Even The First Lady, Melania Trump, was victim to Grok’s deepfakes.

The First Lady Targeted by Grok

Previous misuses of deep fakes led the First Lady to support a bipartisan group of politicians to pass the Take It Down Act, which makes it a federal crime to publish deepfakes of explicit content without consent. She said at a roundtable, “Every young person deserves a safe online space to express themself freely, without the looming threat of exploitation or harm.”

Mrs. Trump’s words are especially reflective because the issue goes beyond celebrity deepfakes. As one young woman, identified only as Evie, recounted how her selfies were used by others on the platform to create sexually explicit imagery about her, “Why is Grok now generating stories about me getting raped now? How is this allowed?" This happened even before the "spicy mode" was released, showing a long-standing issue with the chatbot’s lack of safety filters. 

Elon Musk's Influence on the Update

Despite Grok’s abuses, Musk has consistently positioned Grok as an unfiltered and "rebellious" alternative to other AI models. Following the launch of a previous version of Grok, he responded to a user on X who praised the bot for being the "most uncensored model of its class yet" by declaring, "Grok is the most fun AI in the world!"

More recently, the existence of the "spicy mode" was first announced in a post on X from an xAI employee, Mati Roy, who stated that "Grok Imagine videos have a spicy mode that can do nudity," a post that was later deleted. 

As a result, Grok's approach is in stark contrast to other AI platforms like ChatGPT and Google Gemini, which have much more restrictive content policies. This is especially concerning given the easy accessibility of the chatbot by minors. Musk posted that “kids love using Grok Imagine, because they can speak their ideas and watch the images update immediately." However, this raises questions of the content that children can be exposed to by Grok. Even though Grok has age restrictions, Weatherbed reported, “The age check only appeared once and was laughably easy to bypass, requesting no proof that I was the age I claimed to be.”