Britain Criminalised the Making of the Picture. Denmark Wants the Face Signed Over for Fifty Years.

On 15 January 2026 a minister at the UK Ministry of Justice signed a set of commencement regulations and wrote to a select committee chair to say so. Three weeks later, in England and Wales, drawing a picture became a crime.

The picture is an intimate deepfake of an adult who did not agree to it. Section 138 of the Data (Use and Access) Act 2025 took effect on 6 February 2026, criminalising “the intentional creation of a purported intimate image of an adult (including ‘deepfakes’) without consent or reasonable belief in consent”, Alex Davies-Jones MP told the Justice Committee. The same provisions cover asking somebody else to make one, “regardless of whether the image is then created.”

Read that clause twice. The offence is complete at the request. Nothing has to exist.

That is one of three European answers, arrived at within eighteen months of each other, to a question that used to have no answer at all: who decides what becomes public. London made it a crime committed at the point of manufacture. Copenhagen went looking for something closer to ownership. Brussels declined to treat it as a permission problem and asked instead that the audience be told.

Copenhagen went hunting for a property right and came back with less

Denmark’s move was reported almost everywhere as copyright over your own face. The European Parliament’s own researchers are noticeably cooler about it than the coverage was.

The proposed amendments would require consent from a person before realistic digital imitations of their personal characteristics — appearance and voice among them — are made available, “with the protection lasting for 50 years after the death of that person”, according to an At a Glance briefing published by the European Parliamentary Research Service in January 2026. Caricature, satire, parody and pastiche are excluded in principle. The amendments are expected to take effect in July 2026.

Then the briefing pours cold water on its own subject. “The planned Danish rules do not constitute copyright in the strict sense,” it says, and the Danish authorities “do not intend to designate a new right of copyright for citizens even if the amendments are included in the Copyright Act.”

What they are is a consent requirement with a long tail and a short reach. Enforcement borrows the Digital Services Act’s notice-and-action machinery rather than inventing a remedy, and the rules “would only apply to illegal content on its territory through geo-blocking by very large online platforms and very large search engines.”

So: half a century of protection, stopping at the Danish border, on the big platforms only. Your face keeps a legal opinion about where it appears for fifty years after you stop having one — and loses it entirely somewhere south of Flensburg.

It is also still a draft. The text went to the European Commission for comment through the technical notification procedure, with the window running to 3 February 2026, and the briefing records a view among some experts that extending copyright protection “may bring little added value, and that it may be worth focusing instead on enforcement.”

Brussels is not asking permission. It is asking for a label.

The European approach to the same technology runs on a completely different premise.

From 2 August 2026, deployers must disclose deepfake content “to a natural person upon first exposure at the latest,” and cannot satisfy that by relying on the machine-readable marking a provider embedded in the file, per the European Commission’s guidance on Article 50 of the AI Act. Fines run to 15 million euros or 3% of worldwide turnover. A limited grace period, to 2 December 2026, applies to providers of systems already on the market — not to the disclosure duty itself.

Nothing in any of that turns on whether the person depicted agreed. The obligation regulates what the viewer knows, not what the subject permitted.

One detail inverts where you would expect a law to bite. A person generating deepfakes and posting them on social media in a personal capacity is engaged in “a personal activity” and sits outside the Act entirely; they become a regulated deployer only where the activity is one “through which they gain an economic benefit on a regular basis.” The duty attaches to the business. The hobbyist walks.

Nobody drafting any of this has spent an evening with a streamer

All three instruments treat publication as a switch with two positions. Somebody consents or somebody doesn’t; an image is authorised or it is contraband.

For anyone whose income depends on being on camera, that is not what the decision looks like. It looks like a running negotiation, repriced live, with the audience holding the meter.

Yanlai Wu, Yao Li and Xinning Gui interviewed twenty streamers on the Chinese platform DouYu in October 2020 and published the results in the Proceedings of the ACM on Human-Computer Interaction. What they found was people managing disclosure “based on whether the disclosure would enhance or weaken their attractiveness to the audience,” aware throughout that “disclosing certain personal information would attract audience, which would lead to more monetary income.”

One streamer, a woman who had been doing it full time for five years, put the arithmetic in two sentences. “I didn’t turn on my webcam at first,” she told the researchers. “Then there was no viewer at all.”

The strategies the study catalogues are the tells. Green curtains hung to hide a room. Two monitors so a face can be checked before the audience sees it. Family briefed in advance on when the camera goes on. Seven of the twenty pretended to be single, because viewers tipped less once a streamer was known to be attached — a tactic more common among the women than the men, five to two.

It is twenty people, on one platform, in one country, nearly six years ago, and seventeen of them streamed games rather than anything else. It settles nothing about the wider market. But the mechanism it describes is not confined to DouYu, and it is nowhere in any of the three laws above: control over publication is not a permission granted once. It is a thing people spend their working hours adjusting, in public, at a price the audience sets.

Consent is a moment. Publishing is a job.

Which leaves an odd blind spot at the far end of all this legislative activity. Every instrument here is built for the case where someone’s image escaped them. Almost nothing is built for the far more common case where a person put it out deliberately and would quite like to be found.

The free page is the cleanest version of that. A creator charging nothing has made a decision about what is public and set the price at zero, and unlike consent — which is invisible from outside, which is the whole reason the laws above exist — that decision is legible to anyone looking. It is also the decision the platform files worst. OnlyFans sorts by account name, so someone after girls who post for free on OnlyFans has to sort them elsewhere, working from the pages creators made visible themselves. The honest description of what that produces is narrow: a record of stated prices attached to stated handles. It does not explain why a page is free this month rather than last, it vouches for nobody’s identity, and it contains nothing about material a creator never chose to publish.

Three governments spent eighteen months building machinery for the refusal. The refusal matters, and the Justice Committee correspondence is right that it needed doing.

But the refusal is one decision, taken on the worst day. The rest of it — what goes up, what stays off, what it costs, who gets to see it without paying — is a thousand small permissions, granted and withdrawn by the person in front of the lens, and no statute in Europe has anything to say about those.