Elon Musk’s Grok Sparks Global Backlash Over Non-Consensual Deepfake Images

Even though Elon Musk’s xAI takes a hard-line position in favor of broad free speech, the company has come under scrutiny after reports that its chatbot, Grok, has generated deepfake content involving real people when prompted by users without their consent. This situation exposes a tension between promoting unrestricted expression and the responsibility to prevent unethical or illegal outcomes from AI.

Grok’s Deepfake Capabilities and Controversial Misuse

Grok can produce highly realistic deepfake images within seconds and post them publicly in online threads without asking for permission. Users can trigger the chatbot by tagging it under a photo and issuing prompts that alter a person’s appearance in sexualized ways. Commands such as “put her in a tiny bikini…” or “remove her clothes” have been shared publicly.

The misuse of Grok has drawn criticism from X users. Randi Hipper, a digital asset educator, discovered that a photo of her in gym clothes had been altered to show her in a bikini after another user prompted the chatbot. She described the result as inappropriate and uncomfortable, highlighting the personal impact of such AI-generated content.

To explore Grok’s capabilities, journalist and survivor of child sexual abuse Samantha Smith conducted a test. She uploaded an old photo of herself as a child wearing a communion suit and instructed Grok to create an image of her in a bikini. The chatbot complied, and Samantha reacted with shock, describing it as disturbingly real and deeply unsettling.

Amid growing concerns over the misuse of the AI to generate illegal or exploitative content, Elon Musk posted on X, “anyone using Grok to make illegal content will suffer the same consequences as if they upload illegal content.”

Exploitation of Grok for Adult and Political Content

Meanwhile, some users continue to exploit Grok for adult-oriented content and political messaging, using the chatbot in different ways:

• Adult content creators, including OnlyFans models and erotic performers, have used Grok to boost engagement by encouraging followers to generate images with clothing removed, producing millions of impressions online.
• Others have applied the chatbot for political purposes, such as altering images of public figures. One example involved a user uploading a photo of Donald Trump with Puff Daddy and instructing Grok “remove the pedophile.“

Government Scrutiny and Investigations

Regulators worldwide are beginning to examine the use of Grok more closely. European Commission spokesperson Thomas Regnier stated at a press conference that the authority is “very seriously looking into this matter.” He pointed out that Grok on X allows users to generate sexually explicit material, including images depicting childlike features, and stressed that creating or sharing such content is illegal and has no place in Europe.

In Malaysia, the Communications and Multimedia Commission (MCMC) released a statement on January 3 noting serious concern over public complaints about AI-generated image manipulation on X. The commission said it would “initiate investigations on X users alleged to have violated CMA.”

French authorities also said they will investigate Grok’s use in producing adult-oriented fake images, following reports from numerous women and teenagers of altered photos circulating online.

xAI addressed these concerns on its official X Safety account, stating that it takes “action against illegal content on X, including Child Sexual Abuse Material (CSAM), by removing it, permanently suspending accounts, and working with local governments and law enforcement as necessary.” Additionally, xAI employee Ethan He noted that Grok Imagine has been updated, although it is still unclear whether the changes prevent the generation of illegal or inappropriate sexual content.

A 2023 study by the cybersecurity company Home Security Heroes found that deepfake pornography makes up approximately 98% of all deepfake videos online, with women representing 99% of the subjects, showing the widespread scale of abuse.