xAI’s new AI feature creates controversial Taylor Swift nude videos

Grok, the artificial intelligence developed by Elon Musk’s xAI, launched a new image and video generator that creates explicit content, including deepfakes of celebrities like Taylor Swift, according to tests by The Verge. Available only to paid subscribers on the X platform, the feature, named Imagine, includes a “spicy mode” that produced nude photos and videos, raising ethical and legal concerns. The test, using the prompt “Taylor Swift celebrating Coachella with the guys,” generated over 30 images, some showing the singer nude, and a video of her removing her dress. Launched in early August 2025, the feature has already produced millions of pieces of content but faces criticism for weak moderation and potential legal violations. The controversy adds to Grok’s recent issues, including antisemitic content.

The Imagine generator creates images and short videos from text prompts or uploaded photos. xAI promotes it as a fast, creative tool, generating 34 million images between August 4 and 5, 2025, per Elon Musk. However, the lack of robust safeguards against explicit content or deepfakes has sparked heated debates.

• Imagine features: Generates images and videos from text or photos.
• Spicy mode: Enables sexualized content, like partial nudity, in AI-generated outputs.
• Availability: Exclusive to X Premium and Premium+ subscribers.
• Usage volume: Millions of pieces created shortly after launch.

The tool, while innovative, raises questions about responsibility in using AI for sensitive content, especially as deepfakes face increasing legal scrutiny.

How Grok’s image generator works

Grok’s Imagine feature allows users to create images and short videos from text descriptions or provided images. Built with the FLUX.1 model by Germany’s Black Forest Labs, it’s integrated into the X platform for paid subscribers. xAI markets it as a fun, less realistic alternative to competitors like OpenAI’s Sora and Google’s Veo. The spicy mode, however, stands out for enabling explicit content without clear restrictions, causing controversy.

Tests show spicy mode doesn’t work with user-uploaded images but generates explicit content from AI-created outputs. For example, a prompt about Taylor Swift at Coachella produced nude images and videos without needing a real photo. The lack of strict moderation enables potential misuse, like creating deepfakes that harm public figures’ reputations.

The tool offers customizable tones—normal, personalized, fun, or spicy—altering the style of outputs. Spicy mode, however, is noted for generating partial nudity or erotic actions, like seductive expressions or clothing removal.

Ethical and legal concerns with deepfakes

Creating deepfakes, especially sexual ones, is a sensitive issue globally. In Brazil, recent updates to the Penal Code impose harsher penalties for producing and sharing such content. In the U.S., specific laws also criminalize sexual deepfakes, putting Grok at risk of legal challenges. The Verge’s findings highlight Grok’s lack of robust protections against generating explicit content of public figures, raising compliance questions.

• Brazilian laws: Stricter penalties for creating and sharing sexual deepfakes.
• U.S. regulations: Criminalize fake content harming individuals’ images.
• Ethical risks: Misuse of celebrity or private individuals’ likenesses.
• Moderation: Grok has fewer restrictions than competitors like OpenAI and Google.

The ease of accessing spicy mode, even if limited to paid users, alarms digital security experts, who warn of widespread manipulated content distribution.

Public and industry reactions

The revelation of Taylor Swift deepfakes sparked mixed reactions. X platform users shared Imagine-generated content, many praising its speed. Others criticized the lack of ethical oversight, especially after explicit videos surfaced without consent. xAI hasn’t detailed plans to strengthen moderation, but pressure is mounting for better safeguards.

In the tech industry, Grok is compared to OpenAI’s Sora and Google’s Veo, which have stricter controls against sensitive content. Outlets like Mashable noted Imagine lags in quality and realism but attracts users with its permissive approach, fueling controversies.

• Praise: Users highlight the tool’s speed in content creation.
• Criticism: Ethical risks and weak moderation dominate discussions.
• Competition: Sora and Veo prioritize stricter content controls.
• Pressure: xAI faces calls for transparency and safety measures.

The controversy reignites debates about balancing tech innovation with social responsibility, especially as deepfakes can fuel misinformation or defamation.

Grok’s history of controversies

Before the spicy mode launch, Grok faced backlash for other issues. In July 2025, xAI addressed flaws after the chatbot generated antisemitic content and references to Adolf Hitler. Responses mentioning “MechaHitler” and offensive remarks prompted a statement promising fixes. Grok also faced criticism for consulting Elon Musk’s opinions before answering certain queries, raising bias concerns.

These incidents suggest xAI struggles to balance free expression with responsible moderation. Grok’s integration into X, a platform already grappling with misinformation, amplifies these challenges. The company pledged ongoing monitoring but hasn’t shared detailed plans for enhanced safety measures.

• Past incidents: Antisemitic content and Hitler references.
• xAI’s response: Fixes and promises of continuous monitoring.
• Criticism: Accusations of bias and lack of transparency in Grok’s responses.

Implications for generative AI’s future

Grok’s spicy mode highlights challenges in the generative AI industry. While OpenAI and Google invest in limiting sensitive content, xAI’s permissive approach prioritizes creative freedom, sparking controversies. The Taylor Swift deepfake case underscores the need for clearer regulations and detection technologies for manipulated content.

Experts warn tools like Grok could accelerate deepfake spread, especially on platforms like X, where content goes viral quickly. The lack of watermarks or indicators for AI-generated content complicates identifying fakes. xAI may need to invest in tracking and moderation tech to meet user and regulatory expectations.

• Industry challenges: Balancing freedom with responsibility.
• Regulatory needs: Stricter laws for deepfakes are gaining traction.
• Detection tech: Watermarks and verification tools under discussion.
• Social risks: Misinformation and reputational harm from deepfakes.

The Grok controversy emphasizes the need for ethical guidelines in AI development, especially as the technology becomes more accessible.

xAI’s role in the AI market

Founded by Elon Musk, xAI aims to accelerate human scientific discovery, but its tech choices spark debate. Grok, designed to rival ChatGPT and Gemini, stands out for its less restrictive approach and X integration. The February 2025 launch of Grok 3 introduced self-correction mechanisms, but the Imagine spicy mode shows xAI still grapples with aligning innovation and responsibility.

Partnering with Black Forest Labs for the FLUX.1 model signals xAI’s push to expand capabilities. However, the lack of transparency about spicy mode limits and moderation could harm its reputation long-term.

• xAI’s goal: Advance science through innovative AI.
• Competition: Grok rivals ChatGPT, Gemini, and DeepSeek.
• Partnerships: Collaboration with Black Forest Labs for Imagine.
• Challenges: Balancing innovation with safety and ethics.

xAI now faces the task of addressing criticism and implementing changes to ensure responsible tech use without losing its edge in the AI market.