The European Union is conducting a "serious review" of Elon Musk's AI chatbot Grok after it generated pornographic images involving minors on social platform X.
A European Commission spokesperson, Thomas Renier, stated at a press conference on Monday: "We have noted that platform X or its Grok chatbot has launched a 'Spicy Mode,' which generates content containing explicit pornographic information, with some outputs featuring characteristics of child-like figures." He was referring to the "Spicy Mode" feature launched by Grok last year, which can produce sexually suggestive content. Renier emphasized: "This is by no means 'spicy content'; it is outright illegal activity."
Users of platform X have recently frequently utilized Grok's functionality to perform "digital undressing" on uploaded photos — the subjects of these photos are mostly women, and after processing, the individuals are depicted wearing only underwear or bikinis. The proliferation of such images on this popular social platform has raised alarms among global regulators and cybersecurity advocates, with officials from countries including India, the UK, and France condemning the related posts.
XAI Corp., the operator of platform X and Grok, has not yet responded to requests for comment. Musk posted on platform X on Sunday that the platform would take measures to clean up illegal content, permanently ban relevant accounts, and cooperate with official investigations when necessary. He stated in the post: "Any user who utilizes Grok to generate illegal content will face consequences equivalent to those for directly uploading illegal material."
While most mainstream AI models explicitly prohibit the generation of pornographic images and videos, XAI Corp. has positioned Grok as a "more lenient and open" tool. The system allows the generation of partially nude adult images and sexually suggestive content but explicitly forbids the creation of explicit pornography based on real individuals, as well as any sexual content involving minors. In some countries, such as the United States and the UK, publishing AI-generated intimate "deepfake" images without the subject's consent is itself illegal. The ability to clearly define and strictly enforce these prohibitions has become a key test for the robustness of safety measures in image-generating AI tools.
XAI Corp. has evidently failed to establish effective protective barriers for Grok, a point that has drawn unanimous condemnation from global regulators.
Dutch lawmaker Jeroen Lijnarts pointed out in a statement on Tuesday: "If an AI platform insists on allowing the generation of pornographic content, it must first establish robust, effective, and independently verifiable protective measures." Lijnarts was a key figure in pushing the European Parliament to pass legislation last year that criminalized the generation of AI-created child abuse images.
Lijnarts further added that practices like Grok's, which involve deleting related posts only after the fact, "do not constitute a solution at all," because "the harm suffered by the victims has already occurred and is irreversible."
He emphasized: "If an AI platform chooses to enable the generation of pornographic content, it must first establish robust, effective, and independently verifiable protective mechanisms. Cleaning up after child pornography has been generated is not a solution — once harm is done, it cannot be reversed. AI platforms should bear greater responsibility for protecting children's online safety, but it is evident that the weight of commercial interests seems to be consistently outweighing the work of child online protection."
UK media regulator Ofcom stated on Monday that it was aware of "serious concerns" regarding Grok's functionality and had "urgently contacted platform X and XAI Corp., requesting them to specify the concrete measures taken to fulfill their legal obligations for protecting UK users."
The French government alleged last Friday that Grok generated "clearly illegal" pornographic content on platform X without the subjects' consent, suggesting this matter may violate the EU's Digital Services Act. This act requires large platforms to take measures to reduce the risk of illegal content dissemination.
India's Ministry of Electronics and Information Technology has ordered a comprehensive review of Grok's safety features; Malaysian authorities stated that they are investigating the matter after receiving complaints about Grok generating "indecent content."
Musk's platform X was previously already under investigation for涉嫌 violating the EU's Digital Services Act and was fined 120 million euros (approximately $140 million) in December last year for non-compliance — the first penalty issued under this controversial content governance legislation. The EU's key regulatory actions against US tech companies like X have faced strong criticism from the former US Trump administration, which claimed that the European regulators' approach essentially amounts to "censoring free speech."