The Investigations of Grok AI on X
4.4 million images were posted by Grok A.I., with around 1.8 million of them being nonconsensual sexualized images of women. The social media platform X (formerly Twitter) is currently under investigation after its AI, Grok, has been used to create fake sexual images of unconsenting people, including underage children. Grok AI, created by Elon Musk and built into X for users to access, gained the ability to alter images on the platform, including those posted by other users. Notably, two years ago, X made Grok accessible to all users for free, enabling an even wider range of people to use the AI and its features. This has led to an abuse of the technology to create pornographic imagery of user photos, including minors. In response, several countries worldwide have banned or are investigating X, including the European Commission. Thomas Regnier, an EU digital affairs spokesperson, had this to say about the situation: “Grok is now offering a ‘spicy mode’ showing explicit sexual content with some output generated with childlike images. This is not spicy. This is illegal. This is appalling,” highlighting the severity of the matter to officials across the world.
X’s Response
X responded to this situation in a post on January 6th, claiming that it has acted against the illegal content being created on X, which included “child sexual abuse material” through “removing it, permanently suspending accounts, and working with local governments and law enforcement as necessary.” and warned that anybody who uses Grok to create this content will suffer similar consequences to those who upload illegal content. The European Commission also fined the platform 140 million dollars for “violating the EU’s digital content advertising transparency rules” and also for its lack of proper user verification methods. This is the first penalty under the EU’s Digital Services ACT (DSA). The European Commission gave X 60 working days to elaborate on how they intend to strengthen the “user vetting processes” and 90 days to develop a plan for cleaning up the “advertising repository” and handling “data access-related infringements”. The European Commission’s Executive Vice President for tech sovereignty, security, and democracy, Henna Virkkunen, stated that “With the DSA’s first non-compliance decision, we are holding X responsible for undermining users’ rights and evading accountability.”
CBS News also recently prompted the Grok AI tool about the situation and its ability to generate sexualized images of unconsenting individuals, and Grok itself admitted that it “should face meaningful regulation” and explained how a loophole through tricking the AI into treating images of real people as fictional was utilized to generate these images.
The United States Response
Earlier this month, amid multiple investigations by the European Commission, certain U.S. states also took action. California’s Attorney General Rob Bonta began an investigation into X, while Senator Ted Cruz of Texas introduced the “Take it Down” Act. While X additionally commented that it has: “implemented technological measures to prevent the [@]Grok account on X globally from allowing the editing of images of real people in revealing clothing such as bikinis. This restriction applies to all users, including paid subscribers.”
However, on February 4th, the French Police raided X’s Paris offices to investigate potential complicity by X in this situation. Also, they interviewed Elon Musk, former X CEO Linda Yaccarino, and their employees as witnesses. While X’s response seems optimistic, the numbers offer a contrasting view of the potential damage already done by the A.I.
(CBS News with Grok AI)
The Reported Numbers
(New York Times figure of image generation by Grok)
According to the New York Times, within nine days, the AI Grok posted over 4.4 million images, and at least 1.8 million of them were sexualized images of women. The Center of Countering Digital Hate estimated that over three million images contained sexual imagery of men, women, and children. The Times found that the amount of nonconsensual images surpassed that of deepfake content (realistic A.I.-Generated imagery) from other websites, with Imran Ahmed, the chief executive of the Center for Countering Digital Hate, adding that, “This is industrial-scale abuse of women and girls”. They argued that despite these “nudifying” tools already existing, the larger problem is that these tools have never been so easy to access and distribute on such a massive platform like X. The Center for Countering Digital Hate collected a sample of 20,000 images posted by Grok from December 29 to January 8th with 101 sexual images of children, and estimated that there were more than 23,000 images of children. Comparatively, The Times stated that “One of the largest forums dedicated to making fake images of real people, Mr. Deepfakes,” contained 43,000 sexual deepfake videos of 3,800 at “the peak of its popularity in 2023,” based on data from Stanford and the University of San Diego.
The Potential Repercussions
While it’s currently unclear how all of this may affect several factors, such as AI use and application, as well as global social media laws, X claims to be moving forward with regulating Grok and improving its safety during these investigations. However, with several countries banning social media use for teenagers and younger, this incident may prompt additional countries to follow suit or enact stricter social media laws globally. As for the A.I. implications, this situation on X could prompt lawmakers to regulate A.I. further, but given the United States’ current investment across a multitude of companies, it still seems unlikely to hamper A.I. development. At the very least, it will hopefully result in companies taking more precautions with their A.I. bots.
