Elon Musk’s xAI has come under fire after its new video-generating tool, Grok Imagine, reportedly created an explicit deepfake video of pop star Taylor Swift.

Launched this week for Premium+ subscribers on X, Grok Imagine allows users to generate 15-second videos from image or text prompts, including with a ‘Spicy’ mode that permits NSFW content.

While some prompts are filtered, a report by The Verge claims the tool was able to produce a topless clip of Swift without explicitly requesting nudity – simply by asking it to show her “celebrating at a music festival.”

The generated six-second clip, blurred by The Verge before publication, showed Swift removing a dress to reveal her breasts.

It has sparked outrage online, with critics accusing xAI of enabling non-consensual deepfake pornography.

Despite user guidelines prohibiting pornographic depictions of real individuals, Grok Imagine’s filters appear easily bypassed.

Industry insiders have called the rollout reckless, with The Verge describing it as “a lawsuit waiting to happen.”

This isn’t the first time Swift has been the target of AI exploitation.

In early 2024, explicit AI-generated images of her went viral on X and other platforms, prompting widespread condemnation and urgent calls for legislative reform.

US lawmakers have since proposed bipartisan legislation to criminalise the creation and distribution of deepfake pornography.

xAI has not commented on the Swift video, and neither Musk nor Swift’s representatives have responded publicly.

However, Musk did post on X that user activity around Grok Imagine was “growing like wildfire,” with over 34 million images created since launch.

The controversy puts a spotlight on AI content moderation, especially as Grok Imagine competes with tools from OpenAI, Google, and Runway.

While Grok’s visual quality still suffers from the “uncanny valley” effect, its capability to rapidly generate stylised videos from user prompts, including controversial ones, raises serious ethical and legal red flags.