xAI has responded to a new investigation accusing Elon Musk’s AI chatbot Grok of continuing to generate non-consensual explicit sexual deepfakes. In a direct statement posted on X (formerly Twitter), the company’s official Safety account outlined strict prohibitions and clarified the layered defences already in place to block misuse.
The statement came hours after NBC News published its report, claiming that Grok can still generate sexually explicit images of people without consent. This comes despite the company’s earlier pledge to stop it from happening, utilising stricter safeguards.
X’s Safety account posted the response, stating, “We strictly prohibit users from generating non-consensual explicit deepfakes and from using our tools to undress real people. xAI has extensive safeguards in place to prevent such misuse, such as continuous monitoring of public usage, analysis of evasion attempts in real time, frequent model updates, prompt filters, and additional safeguards.”
Grok accused of generating explicit deepfakes
In a previous investigation report from NBC News, it was found that despite xAI’s earlier public pledge to halt abusive deepfakes following backlash and government scrutiny, Grok’s image-generation tool (Grok Imagine) is still being used to create sexual depfake images of real people without consent. The outlet said that its review identified dozens of AI-generated sexual images and videos depicting real individuals that were posted publicly on X over the past month.
The report clarified that neither X, nor Musk initially responded to requests for comment. It was only after the story was published, X’s Safety account issued the statement. The original report framed the findings as evidence that Grok continues to enable “sexual deepfakes” months after xAI promised stronger protections.
xAI’s clarifies its stand
xAI’s response, however, confirmed that its safeguards are proactively enforced and the firm didn’t deny the model’s capabilities to generate so. The company clarified that its real-time monitoring, prompt filtering, and rapid model updates are key tools in its defence strategy against deepfakes. It reiterated a zero-tolerance policy targeting non-consensual explicit content and “undress” requests involving real people.
Grok’s image generator has faced repeated criticism and regulatory attention in recent months, including investigations by state attorneys general, congressional inquiries, and international probes over concerns about deepfake pornography and child safety. xAI has maintained that it actively removes violative content and reports serious cases to law enforcement.
Industry experts have stated that while most other major AI providers, like Google and OpenAI, have imposed blanket bans on explicit image generation, even of consenting adults, xAI positioned Grok as more permissive, aiming for fewer restrictions while still enforcing rules against non-consensual use.