Brussels has never met a technological problem it did not wish to regulate first and understand later.
From cookie banners that irritate without enlightening, to digital laws so intricate that even their authors struggle to explain them, the European Union’s instinct is invariably the same: legislate early, often and expansively. It is therefore tempting to greet the European Commission’s investigation into Elon Musk’s AI chatbot Grok with a weary sigh. Another probe, another press release, another assertion of regulatory muscle.
And yet this case resists easy cynicism.
The EU’s decision to scrutinise Grok over the proliferation of sexual deepfake imagery on X is not merely an example of bureaucratic overreach or institutional self-importance. It touches on something far more fundamental: the ability of modern technology to weaponise humiliation, strip individuals of dignity and circulate sexualised images without consent, at a scale previously unimaginable.
For once, the problem Brussels is addressing is not hypothetical.
The emergence of generative AI capable of manipulating images has dramatically lowered the barrier to producing sexually explicit material featuring real people who never consented to its creation. Women, children and public figures have found themselves digitally undressed, altered or placed into explicit scenarios with a few keystrokes. These images do not remain in dark corners of the internet; they travel, rapidly and relentlessly, through mainstream social platforms.
No amount of free-speech absolutism can disguise the reality that non-consensual sexual deepfakes are a form of abuse.
That does not mean, however, that the European Commission should be given a free pass on its methods. The Digital Services Act, under which the investigation into X is being conducted, exemplifies Brussels’ tendency to respond to fast-moving technological change with dense, prescriptive regulation enforced by the threat of colossal fines. The risk is not merely over-compliance, but paralysis: companies retreating from innovation altogether, or deploying blunt restrictions that harm legitimate users while failing to stop determined abusers.
There is also an uncomfortable truth that Brussels rarely acknowledges. European institutions were not blindsided by generative AI; they were simply slow. Deepfake technology did not materialise overnight. It evolved over years, largely ignored by lawmakers until public outrage made inaction politically impossible. The Commission’s sudden urgency now has the feel of an authority scrambling to reassert relevance rather than one calmly guiding technological development.
Elon Musk, for his part, has done little to inspire confidence. His vision of X as a digital town square, lightly moderated and ideologically unrestrained, collides head-on with the realities of AI-driven content creation. The idea that platforms can rely on community standards and post-hoc enforcement when machines can generate thousands of abusive images in minutes is plainly fanciful.
Safeguards introduced after public backlash are welcome, but they raise an obvious question: why were the risks not addressed before deployment?
This is where Brussels’ concerns are legitimate. The failure to conduct serious, pre-emptive risk assessments for AI systems capable of producing sexualised imagery is not a trivial oversight. It speaks to a broader culture within parts of the tech industry that treats social harm as an externality — unfortunate, regrettable, but ultimately someone else’s problem.
Yet the Commission must tread carefully. If this investigation becomes a pretext for asserting sweeping control over AI development, content moderation algorithms and platform design, it will confirm the worst suspicions of critics who see Brussels as addicted to regulation for its own sake. The EU’s credibility depends on demonstrating that it can distinguish between targeted intervention and ideological crusading.
There is a narrow but vital path to be walked here. Non-consensual sexual deepfakes demand firm, focused action: clear obligations on platforms to prevent generation of identifiable sexual imagery without consent, rapid takedown mechanisms, cooperation with law enforcement where criminal thresholds are crossed. These are defensible, proportionate measures rooted in harm prevention rather than moral posturing.
What would be far less defensible is using this moment to expand regulatory authority into every corner of AI-driven communication, treating all generative technology as inherently suspect and all platforms as guilty until proven compliant.
Brussels often speaks of “European values”. If those words are to mean anything, they must include both the protection of individual dignity and a restraint against bureaucratic excess. The Grok investigation will test whether the EU can uphold one without sacrificing the other.
For once, the problem is real. The danger lies not in acting — but in acting as Brussels so often does: too broadly, too rigidly, and too late.
