Grok’s Persistent Problem: Unconsented Deepfakes Continue on Musk’s AI Platform
Months after Elon Musk’s artificial intelligence firm, xAI, pledged to implement stringent restrictions against the creation of harmful sexualized deepfakes, its Grok chatbot reportedly remains a platform for generating and hosting nonconsensual explicit images and videos of women. This concerning revelation emerges as xAI’s parent company, SpaceX, prepares for a monumental IPO, one of the largest in history, casting a shadow over the tech giant’s ethical commitments.
A Disturbing Inventory of AI-Generated Content
A recent analysis by WIRED uncovered a disturbing array of content created and hosted by Grok Imagine, xAI’s generative AI system. This includes images and videos depicting celebrities and even a prominent politician in highly compromising scenarios: held against their will by a giant figure, performing sex acts, and in states of full nudity. While some creations leaned towards animated styles, a significant portion were photorealistic, presenting scenarios that could be mistaken for reality.
WIRED’s investigation, which reviewed hundreds of publicly accessible Grok Imagine links on Grok.com, identified dozens of sexualized AI images and videos, many created without the explicit consent of the subjects. These links were not confined to Grok’s native site; some were subsequently shared on X (formerly Twitter), highlighting the potential for wider dissemination. The public posts, lacking timestamps, likely represent only a fraction of the content being generated, as creations do not appear to be public by default.
Falling Short of Industry Standards
Compared to other mainstream generative AI systems, Grok appears to lag significantly in its safety protocols. Henry Ajder, a leading expert on deepfakes with a decade of experience tracking explicit AI content, notes, “While Grok and X may have made some amendments to their model, particularly following the backlash around nudification at the beginning of the year, they still have not done a sufficient job to bring it up to the standard of the other mainstream tools that are available.”
This isn’t xAI’s first encounter with controversy. Since January, the company has faced a barrage of lawsuits and regulatory scrutiny globally, following an initial wave of “nudification” images generated via Grok on X. Users, predominantly men, exploited the chatbot to alter images of women into “bikinis” or “string bikinis.” Alarmingly, a class-action lawsuit filed in California federal court in March alleged that images of apparent minors were also sexualized.
Promises vs. Reality: xAI’s Safeguards Under Scrutiny
In response to earlier criticisms, xAI asserted that it had implemented safeguards to prevent the creation of nonconsensual and sexualized deepfakes, reiterating a strict ban on child sexual abuse material (CSAM). However, the recent WIRED findings suggest these measures may be insufficient or inconsistently applied.
Neither xAI nor X initially responded to WIRED’s inquiries regarding the explicit content. Interestingly, shortly after WIRED contacted the companies, the identified explicit images and videos on Grok.com became unavailable, and related Grok Imagine links on X were removed for policy violations. An X safety account had previously stated in April, “We strictly prohibit users from generating nonconsensual explicit deepfakes and from using our tools to undress real people,” in response to an NBC report.
High-Profile Targets and Disturbing Prompts
Among the nonconsensual content hosted on Grok.com, WIRED identified images and videos featuring multiple celebrities and even US Representative Alexandria Ocasio-Cortez. These depictions included women scantily clad and held captive in a “giant” man’s fist. One particularly disturbing prompt described a celebrity pleading for release as a giant hand tightens its grip, with the giant man “leaning in and licking her face up and down.”
A comparative test by WIRED revealed the stark difference in safety guardrails: prompts that generated illicit content on Grok were rejected as inappropriate by OpenAI’s ChatGPT, Meta AI, and Anthropic’s Claude. Google’s Gemini, while rejecting one prompt, did generate an image of a celebrity held by a giant hand for another.
Legal Ramifications and a “Historic” Development
The issue extends to personal legal battles. A Grok Imagine video, also shared on X, appeared to depict Ashley St. Clair, a former partner of Elon Musk and mother to one of his children, dancing in a bikini. St. Clair initiated legal action against xAI in January after similar sexualized deepfakes of her surfaced on X. The post was subsequently removed after WIRED’s contact.
Carrie Goldberg, St. Clair’s attorney and a vocal advocate against nonconsensual sexualized imagery, emphasizes the unprecedented nature of this situation. While the creation of sexual abuse material featuring celebrities isn’t new, the ability to instantly generate and publish such images on social media represents a “historic” shift. Goldberg highlights, “So there was instant publication of the digital nude images. And not only that, but the world’s richest man was promoting the nudification technology to his 240 million followers on X, monetizing it, and doing so on a platform that is fo”
For more details, visit our website.
Source: Link
Leave a comment