Spain's conservative People's Party published an AI-generated campaign video this week that depicts the prime minister's wife, a former Socialist president, and a former party member appearing to confess to corruption, nepotism, and judicial interference. The video carried a tiny, near-transparent AI disclaimer at the bottom of the frame. By Friday it had nearly 2 million views across the PP's official social media channels. The deepfakes are specific and pointed. An AI-generated Begoña Gómez discusses receiving a university post through her husband's position. A synthetic José Luis Rodríguez Zapatero talks about hiding jewellery. Former Socialist member Leire Díez appears to describe undermining judicial and law enforcement institutions. Each line references real ongoing investigations in which the actual individuals deny wrongdoing — the video launders accusation through satire. Legal response has been swift but structurally weak. Díez filed a legal challenge accusing the PP of defamation and digital identity theft. But a majority of nine legal experts consulted by El País concluded the video was likely legal, given the AI label and satirical framing. Criminal law professor Juan Antonio Lascuraín drew the distinction cleanly: legality is one question; whether a party aspiring to govern should fabricate confessions from its opponents is another entirely. The PP's defence is revealing. Spokesperson Borja Sémper argued there was no malicious intent because "everyone can see" it's AI-generated — a defence that simultaneously claims the video is obviously fake and relies on it being persuasive enough to rack up millions of views. Digital transformation minister Óscar López countered that audiences unfamiliar with AI could easily mistake the content for reality. This is not an isolated phenomenon. A Wesleyan University research group identified over 200 AI-generated or AI-enhanced political ads in the US 2026 midterm cycle, overwhelmingly tied to Republicans, up from under $50,000 in total spending on such ads in 2024. The infrastructure for synthetic political speech is scaling faster than any regulatory or legal response. The Spanish case is particularly instructive because of what it normalises. When a major party — currently leading in polls and expected to win the 29 November general election — deploys deepfakes on its official channels, the tactic is no longer fringe. Associate professor José Javier Olivas called it a "red line" in campaign conduct, noting it qualifies as disinformation regardless of the satirical packaging. The structural question is who bears the cost of this normalisation. Not the PP, which gets millions of views and faces no meaningful legal consequence. Not the platforms hosting the content. The cost falls on the individuals whose likenesses were stolen, voters who must now parse every political video for authenticity, and the baseline level of trust required for democratic discourse to function at all.
