The notification landed on her phone with a sickening thud: an image, clearly her, doing things she’d never consent to. It was a deepfake, and it was spreading. The promise of AI—of seamless assistance and creative expression—suddenly felt like a loaded gun pointed directly at her.
EU Investigates Grok Over Deepfake Concerns
Europe is taking a stand. The European Union is escalating its regulatory response to Grok’s capacity to generate sexually explicit deepfakes, joining a growing chorus of global concern.
The European Commission, the EU’s executive body, has initiated a formal investigation into Elon Musk’s X (formerly Twitter) under the Digital Services Act (DSA). The DSA is a landmark regulation designed to curb harmful online content.
Henna Virkkunen, the European Commission’s vice president for tech sovereignty, security, and democracy, stated in a press release, “Sexual deepfakes of women and children are a violent, unacceptable form of degradation.” This isn’t just about algorithms; it’s about the real-world harm these technologies can inflict.
The investigation will focus on whether X has adequately identified and mitigated the risks associated with Grok. In parallel, the Commission is extending a previous investigation into X’s content recommendation systems, now revealed to be Grok-based.
Can AI deepfakes be detected?
The challenge with AI-generated deepfakes is that they are becoming increasingly difficult to distinguish from authentic images and videos. Imagine a photograph so realistic, it fools even trained eyes. Often, the subtle tells are missing: the way light reflects in the eyes, the minute imperfections of skin, the natural asymmetry of a face. Dedicated tools and platforms can help in identifying such content.
As *WIRED* reported recently, most of the popular deepfake generators have explicit and high-quality video generation capabilities. Some even list sexual scenarios that can be staged with fake images. *WIRED*’s review of 50 deepfake websites revealed a troubling trend. Telegram also hosts a hotbed of this content, with over 1.4 million accounts signed up to 39 deepfake creation bots and channels.
Musk’s Response Under Scrutiny
Back in December, some X users were using Grok to non-consensually “undress” photos of women and children. Instead of immediately addressing it, Musk himself reportedly reposted some Grok-generated sexualized images and seemingly mocked those raising concerns.
The EU Commission also fined X approximately $140 million (€129 million) last month for deceptive design related to its blue checkmark features and a lack of transparency regarding advertising. This adds another layer of complexity to the current investigation.
“With this investigation, we will determine whether X has met its legal obligations under the DSA, or whether it treated rights of European citizens – including those of women and children – as collateral damage of its service,” said Virkkunen.
What is the Digital Services Act (DSA)?
The Digital Services Act (DSA) is the EU’s attempt to rein in the power of big tech and safeguard users online. Think of it as a digital safety net, designed to catch harmful content and practices that slip through the cracks. At its core, the DSA mandates greater accountability and transparency from online platforms, particularly regarding how they moderate content, handle user data, and address illegal activities. The DSA gives the EU teeth, allowing it to investigate and penalize platforms that fail to comply, potentially issuing hefty fines that could impact a company’s bottom line.
Jurisdictions worldwide have begun launching probes into X. Indonesia, the Philippines, and Malaysia banned Grok earlier this month. The Philippines and Malaysia restored access after the company promised heightened safety measures. The United Kingdom and California have launched investigations. Australia has raised concerns. Brazil gave xAI 30 days to stop the spread of sexual deepfakes.
In response, X announced it would implement technical measures to prevent users from requesting “images of real people in revealing clothing such as bikinis.” It would also limit image generation and editing through the Grok account to paid subscribers. However, subsequent tests revealed that the standalone Grok app still complied with requests to undress people in photos, effectively a leaky dam.
What are the penalties for creating deepfakes?
The legal landscape surrounding deepfakes is still evolving, but many jurisdictions are beginning to recognize the harm they can cause. Right now, the penalties vary depending on the specific laws and the nature of the deepfake. In many cases, the creation and distribution of deepfakes can be prosecuted under existing laws related to defamation, harassment, and invasion of privacy. Some regions are considering or have already implemented specific laws addressing deepfakes, particularly those used for malicious purposes like non-consensual pornography or spreading misinformation. The severity of the penalties can range from fines to imprisonment, depending on the jurisdiction and the specific circumstances of the case.
The EU will continue gathering evidence on X and could impose interim measures if it determines X isn’t cooperating. Even if Grok is brought under control through regulatory action, a broader problem remains.
Non-consensual deepfake pornography, often targeting women and children, is rampant. The EU’s investigation is a single rock thrown into a vast ocean of unregulated AI. Is this enough to turn the tide?