X's AI Deepfake Scandal Is the DSA Compliance Test Dating Apps Will Face Soon

The European Commission's formal investigation into X over its Grok AI chatbot—announced 26th January and centring on whether the platform conducted adequate risk assessments before deployment—has just given every dating app CTO a reason to review their own AI roadmap. At issue: Grok reportedly generated thousands of sexually explicit deepfake images daily, including non-consensual intimate imagery of real individuals, before X imposed restrictions. The Commission is examining whether this constitutes a Digital Services Act (DSA) violation, with potential penalties reaching 6% of X's global revenue.

For dating operators, this isn't distant social media drama. It's the first major regulatory test of how platforms will be held accountable when AI features designed for engagement produce non-consensual intimate content—a risk the dating industry has been quietly managing since generative AI became commercially viable.

Create a free account

Unlock unlimited access and get the weekly briefing delivered to your inbox.

Register free

No spam. No password. We'll send a one-time link to confirm your email.

Why dating apps saw this coming

Most major dating platforms have adopted AI cautiously when it comes to visual content. Bumble (BMBL) uses AI for profile moderation and to detect fake accounts. Match Group (MTCH) properties deploy machine learning for recommendations and fraud detection. Grindr (GRND) has tested AI chat suggestions. But generative image tools? Nearly absent from consumer-facing features.

The calculation was straightforward. Generative AI that produces or manipulates photos of real people introduces consent risks that dwarf the potential engagement upside. Dating apps already battle intimate image abuse—revenge porn, catfishing with stolen photos, blackmail schemes. Adding a tool that could generate convincing fake nudes or manipulate member photos was a risk-reward equation that didn't pencil.

X apparently made a different calculation. According to the Commission's announcement, Grok's image generation capabilities were deployed without adequate assessment of risks including 'manipulated sexually explicit images' and content that could constitute child sexual abuse material.

The platform's scale—X is designated a Very Large Online Platform under the DSA, with over 45 million monthly EU users—meant those risks materialised at volume. Reports suggest thousands of explicit deepfakes were generated daily before restrictions were imposed in January 2025.

The enforcement pattern taking shape

This is X's second significant DSA action in as many months. The Commission issued a €130 million fine in December 2025 over X's paid verification programme, which regulators argued facilitated impersonation and deceptive practices. Owner Elon Musk characterised both actions as attacks on free speech, a defence that sidesteps a key distinction: non-consensual intimate imagery isn't a speech matter in most EU member states. It's illegal.

The Commission's focus on 'gender-based violence' and mental health impacts mirrors language increasingly common in dating app regulatory discussions. The UK Online Safety Act (OSA), which came into force for largest platforms in March 2024, specifically addresses intimate image abuse. The DSA's risk assessment requirements—which mandate platforms evaluate harms before deploying new features—were written with precisely this scenario in mind.

What's notable is the speed. The Commission ordered X to preserve all Grok-related internal documentation through 2026 on 14th January. The formal investigation followed twelve days later. This isn't the multi-year regulatory process the industry became accustomed to during the GDPR era. Enforcement under the DSA is operating on a different timeline, and the Commission appears to be using X as a demonstration case for what happens when platforms deploy first and assess risk later.

For dating operators, the lesson is procedural as much as substantive. The investigation hinges not just on whether harm occurred, but on whether X conducted proper risk assessments before launch.

The Commission's statement specifically questions whether the company 'adequately evaluated and addressed risks' associated with Grok's deployment. Under the DSA's Article 34, Very Large Online Platforms must assess 'systemic risks' including those related to gender-based violence before introducing features that could amplify them.

Dating apps, even those below the VLOP threshold, should assume this standard will cascade. If a platform with X's resources can't claim ignorance about deepfake risks in 2025, a dating app certainly can't in 2026.

The compliance calculus changes

The investigation also expands an existing review of X's content recommendation systems, now incorporating the platform's recent shift to a Grok-powered algorithm for suggesting posts. This matters because it signals regulators are examining AI integration holistically—not just standalone features, but how AI systems interact with existing platform mechanics to amplify risk.

Dating apps use recommendation algorithms extensively. Every swipe queue, every 'people you may like' suggestion, every notification about profile views relies on algorithmic ranking. As those systems incorporate more sophisticated AI—large language models for chat, computer vision for photo analysis, generative features for profile assistance—the compliance question becomes harder to parse. At what point does an 'enhanced' profile photo tool become a deepfake generator? When does an AI chat assistant cross into impersonation?

The industry doesn't have clear answers yet, but the Commission's approach to X suggests the burden of proof will sit with platforms. Demonstrating that you conducted thorough risk assessments, implemented safeguards, and monitored for misuse isn't optional. It's the defence you'll need if your AI feature produces harmful content at scale.

X's case will likely take months to resolve, but the precedent is already forming. Dating apps that have held off on generative AI features now have regulatory cover for that caution. Those considering adding them have a blueprint for what due diligence looks like—and what happens when you skip it. The UK data regulator has also opened its own probe into X, signaling that DSA enforcement is part of a broader regulatory pattern. The Commission has made X the test case. The rest of the industry should be taking notes.