EU deepfake law: the ban the Socialists and Democrats want and the proof it would require

What the S&D draft for an EU deepfake law asks for

The S&D draft asks for an EU law that bans harmful, non-consensual and deceptive deepfakes and makes platforms and AI providers stop their spread, but no bill has been tabled or voted on.

A ban on harmful, non-consensual and deceptive deepfakes

Read together, the draft’s adjectives point at fakes that mislead people and use someone’s likeness without permission, rather than at synthetic media in general. According to POLITICO and CDE News, German S&D lawmaker Tiemo Wölken wants protection to reach beyond artists and cover non-consensual deepfakes that are not a crime under current law.

Platforms and AI providers asked to prevent, not react

Under the S&D draft reported by POLITICO, tech platforms and AI providers would have to prevent deepfakes from spreading proactively, instead of waiting for victims to report them. Today it is usually the person depicted who has to find the fake and flag it. Danish S&D lawmaker Christel Schaldemose asks the European Commission to assess whether an EU law would add value to the national initiatives already under way.

What the draft does not say yet

CDE News points out that the draft does not explain how the ban would be enforced and describes no takedown mechanism, so who would have to do what under an EU deepfake law is still an open question.

What EU law already says about deepfakes

Until there is an EU deepfake law, EU rules do not ban deepfakes as such: since August 2, 2026, the AI Act has required anyone who uses an AI system to generate or manipulate image, audio or video content that constitutes a deep fake to disclose that the content is artificial. Labels flag what is artificial; TrueScreen protects the other side, certifying with a forensic methodology the authenticity of content acquired at the source, while the disclosure duty stays with whoever uses the AI system.

Current EU deepfake regulation relies on disclosure and on procedures for illegal content; protecting face and voice is something the Commission says it will examine.

AI Act: the duty to disclose deepfakes (Article 50)

The duty under Article 50(4) of the AI Act falls on whoever uses the AI system; for evidently artistic or creative works, the disclosure must not hamper the enjoyment of the work. Our analysis of the transparency obligation under Article 50 covers who has to label what.

The nudification ban in the AI omnibus package

According to the European Parliament’s final vote of June 16, 2026, the omnibus package adds to the AI Act a ban on systems that generate sexually explicit images of identifiable people without consent, with application expected from December 2, 2026, as Eunews reported. The measure is aimed at nudifier apps.

DSA: notice and action on illegal content

The Digital Services Act requires hosting services, including online platforms, to provide a notice and action mechanism for illegal content (Article 16). A non-consensual fake that is not illegal under national law may fall outside the mechanism, and a notice does not guarantee removal.

Face and voice: what Parliament asked and what the Commission answered

Article 50 imposes a transparency obligation on whoever uses the system; it does not introduce a harmonized right, specific to face and voice, for the person depicted. On March 10, 2026, Parliament adopted a non-legislative resolution on copyright and generative AI (2025/2058(INI), rapporteur Axel Voss). In its official follow-up on paragraph 27, the Commission says it “will examine the issue within the context of copyright, personality rights or the AI policy frameworks and explore solutions to curb the deceptive use of deepfake technologies.” Its consultation on a targeted copyright initiative, which feeds into the review of Directive (EU) 2019/790, closed on June 25, 2026.

Why member states are writing their own deepfake laws

Without an EU deepfake law, member states are filling the gap one at a time, and the differences between their choices sit at the center of the S&D argument.

Denmark, Cyprus and Belgium: copyright over face and voice

Denmark introduced a bill in the Folketing on October 8, 2026, Euronews reported; it has not been passed, and the government hopes it will take effect on January 1, 2027. According to POLITICO, the bill amends copyright law against the non-consensual sharing of digital replicas of a person’s face, voice and other characteristics, with exceptions for parody and satire. POLITICO also reports that Cyprus amended its copyright law in March, Belgium proposed a neighboring right in September, the Netherlands is running a consultation and Germany has separate measures. Others look elsewhere, for instance at protecting a voice with a sound mark.

The risk of a different rule in every country

Wölken argues that “separate national approaches create legal confusion that bad actors could exploit,” as quoted by POLITICO and CDE News. Leentje Grillaert, a co-author of the Belgian proposal, told the same outlets that “the EU is heading in the right direction but needs to move faster.” Picture a fake video of a chief executive circulating across the EU: without an EU deepfake law, it could fall under copyright in one country, under a different regime in another and under none in a third.

Why a deepfake ban needs proof of the original

Banning deceptive deepfakes means telling them apart from genuine content that someone disputes. A person targeted has to document the fake before it is removed or altered, and a person imitated has to be able to show the original. A file of unknown origin produces an opinion; a file sealed at capture produces a defensible result. This is where TrueScreen leaves little room for disputes: it acquires the original in a controlled environment, verifies its data and documents every step in a technical report an expert can review.

An EU deepfake law against deceptive fakes would work only if someone can show what the original was and when it was captured.

Telling a deceptive fake from a genuine recording someone disputes

A ban cuts both ways: it also makes “that video is a deepfake” a more tempting defense for anyone caught on a real recording. Whoever relies on the genuine file has to explain where it came from and whether it has changed, so it pays to know how to demonstrate content authenticity before a dispute starts.

Detection is a signal, not proof

Deepfake detection estimates how likely it is that a file was manipulated; it does not establish how and when the file was captured, and it can be wrong. TrueScreen combines both defenses: it certifies at the source what it acquires and, for files it did not acquire, also offers detection of deepfakes and AI-generated content, as one more probabilistic signal. It covers photos, videos and audio, not text, and is available on all plans through the Media Inspector in the Forensic Browser. Our guide on how to detect a deepfake explains what to look for.

People targeted need to document before removal

Posts get deleted and accounts get renamed. In a hypothetical case, a professional finds a video in which she appears to say things she never said, posted from an account that may soon disappear; before reporting it, she captures the post, the profile page and the URL where it circulates. If the content is intimate in nature, do not download or share it: contact a lawyer and the competent authorities right away, and they will tell you what to keep and how. Our article on deepfakes as evidence in court covers what comes next.

Feature

Capture, analyse and certify any online content.

TrueScreen captures the posts and pages where disputed content circulates, with date and integrity third parties can verify, through its Forensic Browser.

Discover more →

How do you certify original content before someone disputes it?

TrueScreen secures originals before a fake circulates: they are certified with the TrueScreen App, and its Forensic Browser captures the web pages where the fake later spreads. The forensic methodology documents every step of the capture, and the certification lets third parties verify date and integrity. TrueScreen guarantees the how, not the what: it certifies how and when content was captured, not that the recorded scene is true.

Whatever form an EU deepfake law takes, certifying originals before a dispute and capturing the fake where it circulates gives your lawyer a record of both, with a date and integrity that can be checked.

Certifying originals at the source with the TrueScreen App

The TrueScreen App certifies photos, videos, audio and the device screen, through screen recording, at the moment of capture, so a communications team can certify an executive’s statement when it is made, not when it is challenged. Behind that certified statement is a patented forensic methodology: TrueScreen acquires the content at the source, verifies it and documents every step in a technical report. A qualified electronic seal and a qualified timestamp, whose legal basis is the eIDAS Regulation (Regulation (EU) No 910/2014), close the certification.

Capturing the deepfake online with the Forensic Browser

TrueScreen’s Forensic Browser captures the post, the profile page and the URL where a fake circulates, and for each captured page it delivers the page image (original PNG), the source code and the web archive (MHTML), with a manifest of fingerprints. Pages can include third-party data, such as comments and profiles: keep only what you need, on your lawyer’s advice. In a hypothetical case, a fake video of a company’s CEO spreads on social media; the legal team captures the post and the profile page and sets them next to the original video the company had certified with the TrueScreen App.

What TrueScreen certifies and what it doesn’t

TrueScreen guarantees the how, not the what: a staged scene is certified as an authentic recording of a staged scene. The result is certified evidence whose weight the judge decides; like any evidence it can be challenged, and no certification can promise the outcome of a case.

1. 1

   Certify your originals

   Use the TrueScreen App to certify any file you may later need to set against a fake.

2. 2

   Capture the page and the fake

   Capture the post, the profile page and the content in the Forensic Browser before anyone removes or edits them.

3. 3

   Keep the evidence and share it with your lawyer

   Store the certified files and reports, and share them with your lawyer, who decides how to use them.

Conclusion