Spain PP deepfake election ad sparks legal complaint

In plain words

Spain's main opposition party posted a video on its official accounts in which AI-made copies of people linked to the governing Socialists seem to admit to crimes. The video carries a tiny note saying it was made with AI. Critics say many viewers could still take it as real. Treat any political video showing a named person confessing as unverified until a reliable outlet or the person's own channels confirm it, and look for the AI label.

What happened

The Guardian reports that the People's party (PP) launched the ad this week across its social media channels. It used AI to manipulate the likenesses of various people linked to Pedro Sánchez's Socialist party, who appeared to admit to acts such as soliciting prostitutes and taking kickbacks. An AI version of Sánchez's wife, Begoña Gómez, speaks about receiving a university post thanks to her husband's job. A deepfake of former president José Luis Rodríguez Zapatero speaks about hiding jewellery in a safe. The Guardian says both lines nod to ongoing investigations in which Gómez and Zapatero have denied wrongdoing.

By Friday the video had nearly 2m views. Digital transformation minister Óscar López said it was "telling people a colossal lie"⁠¹. The Guardian says the AI disclosure sat at the bottom in tiny, nearly transparent font. Leire Díez, a former Socialist member whose likeness was also used, reportedly filed a court complaint alleging defamation and digital identity theft and asking for immediate removal. General elections are on 29 November.

What is confirmed and what is not

Confirmed by the Guardian's reporting: the ad exists, it was posted on official PP channels, it used AI likenesses, it had nearly 2m views, and a complaint was filed. The complaint's contents are described as "reportedly", so they rest on secondhand reporting.

Claimed: the Socialist side says the video is a lie and easily mistaken for reality. The PP's Borja Sémper says there was no intent to make people believe the characters are real and that the video highlights Socialist hypocrisy. Most of the nine legal experts consulted by El País reportedly think the ad was not illegal because it was labelled and satirical in tone.

Unknown: whether a court will order removal, whether any viewer was actually misled, and how the video will affect voting. The sources give no measurement of viewer belief.

1. Spanish voters seeing the clip in social feeds
2. Politicians and people whose likeness was used
3. Spanish courts and legal system handling the complaint
4. Political parties running campaigns elsewhere, including the US
5. Social media platforms hosting political ads

Analysis

Voters may mistake a labelled fake for a real confession

The most direct consequence falls on voters who see the clip in a feed without context. The mechanism is simple: a realistic synthetic face speaking a confession is processed as evidence, while a faint disclosure at the bottom may go unread. López made this point, saying it is easy for anyone not used to AI to confuse the video with reality. The sources show the label was small and show the view count, but they do not show how many viewers misread it. That link is asserted by the minister, not measured. A sceptical expert would ask for survey or comment data on belief. If the clip is cut and reshared without the label, the risk would grow, though the sources do not report such resharing.

Law and policy: legal and ethical tests diverge

The legal question is contested. According to the Guardian, Díez's complaint invokes defamation and digital identity theft. To succeed, a claimant would generally need to show that the content damaged her reputation or misused her identity in a way the law covers. The experts cited by El País reportedly lean towards the view that the labelling and satirical tone defeat that. Íñigo Navarro said everyone knows those people did not say those things. This is a general reading of the argument, and the sources do not describe the legal tests in detail. The remedy sought in the one complaint described is immediate takedown. Other targets, such as Gómez and Zapatero, are not reported to have filed anything.

Juan Antonio Lascuraín separates legality from whether the ad is appropriate or ethically correct. That distinction matters: if a labelled deepfake is lawful, then the safeguard against it is mostly the label itself, and the sources suggest the label here was weak. José Javier Olivas goes further, calling it a "red line"⁠¹ and saying it qualifies as disinformation. Parties differ: the minister and the academic see harm, the PP sees satire, and the lawyers split between lawfulness and propriety.

Politics and daily life: a precedent for official party use

The notable feature is that a mainstream party used the technique on its official accounts, not an anonymous account. If that is tolerated, TOSWO Research's reading is that other parties could treat it as acceptable, because the cost appears to be criticism rather than a ruling. That is a conditional reading and the sources do not show it happening. The Guardian also notes that in the US a Wesleyan Media Project report identified more than 200 AI-generated or enhanced political ads, most tied to Republicans, and that much AI use carries no disclaimer. Erika Franklin Fowler compared this with less than $50,000 in AI ad spending in the 2024 cycle. These figures concern US ads and do not show that Spain's case is typical, but they indicate the practice is not isolated.

Technology and industry: labels as the main safeguard

The sources name no technical safeguard, such as watermarking or platform detection, and do not say whether the platforms acted. The only protection described is the on-screen disclosure. Its weakness is the gap between meeting a disclosure duty on paper and being noticed by a viewer. The sources do not say whether Spain has a rule on label size or visibility, so that remains unknown.

The strongest reason this reading could be wrong

The strongest counterargument is the PP's and the lawyers': the clips echo known investigations and the satirical tone is obvious, so few people may be deceived and the harm may be mainly to taste, not to belief. Evidence that would change our view includes audience studies showing low confusion, or a court ruling that the label was sufficient. Evidence of widespread belief or unlabelled reshares would strengthen the concern. Only one outlet's report is the basis here, with El País's expert survey relayed second hand, so claims should be kept to what it supports.

What would settle it

The court filing by Díez and any judicial decision on takedown would show how a judge treats the label and the satire defence. The El País article with the nine experts would show the reasoning in full. The video itself, with the label as displayed, lets readers judge visibility. Any statement from the social platforms, and the Wesleyan Media Project report, would add evidence on moderation and scale.

What can be done

Readers can check the label and the source before sharing a clip of a named person confessing, and wait for confirmation from established outlets. Parties could display disclosures in readable size and for the whole video. Regulators could clarify whether such labels suffice, though the sources do not say whether any is planning to. For wider context, see our earlier Signal on the EU ban on nudifier deepfake apps, our incident record since 2018 and how TOSWO measures AI risk.

Questions readers ask

Was the Spanish deepfake election video illegal?

That is disputed. The Guardian says a majority of nine legal experts consulted by El País thought it was not illegal, given the AI label and the satirical tone. A former Socialist member has reportedly filed a court complaint alleging defamation and digital identity theft, and no ruling is reported.

Who is shown in the video?

The Guardian says it shows AI likenesses of people linked to the Socialist party, including Begoña Gómez and José Luis Rodríguez Zapatero. Leire Díez, a former Socialist member, says her likeness was also used.

Did the video say it was made with AI?

Yes, according to the Guardian, but in tiny, nearly transparent font at the bottom. The minister López argues that this makes confusion easy for people unused to AI.

How widely was it seen?

The Guardian reports nearly 2m views by Friday. It gives no data on how many viewers believed it.