Even the voices of the dead can be conscripted into service of present-day agendas
En un momento en que Perú y Colombia disputaban la soberanía sobre la isla Santa Rosa, una voz del pasado pareció alzarse para dar respuestas: la del expresidente Alan García, fallecido en 2019, supuestamente prediciendo el conflicto desde una conferencia universitaria. Pero esa voz era una ilusión fabricada por inteligencia artificial, superpuesta sobre una clase magistral de 2003 que jamás abordó el tema. El incidente nos recuerda que en la era digital, incluso los muertos pueden ser convocados para servir agendas del presente, y que la desinformación encuentra su terreno más fértil precisamente cuando la incertidumbre es mayor.
- Un video falso atribuido a Alan García acumuló 1.500 millones de visualizaciones en TikTok mientras la tensión diplomática entre Perú y Colombia alcanzaba su punto más álgido.
- Periodistas y figuras públicas de renombre amplificaron el contenido sin verificarlo, otorgándole una credibilidad que aceleró su propagación masiva.
- La tecnología de síntesis de voz por IA replicó con tal precisión los rasgos fonológicos de García que millones de usuarios no detectaron el engaño durante una crisis diplomática activa.
- Los verificadores de La República rastrearon el video original —una conferencia de tres horas del año 2003— y confirmaron que García nunca mencionó la isla Santa Rosa ni ningún conflicto con Colombia.
- La organización juvenil del partido APRA emitió un comunicado oficial denunciando el audio como una creación artificial, pero la corrección llegó cuando la narrativa falsa ya había moldeado la percepción pública.
A mediados de agosto, mientras Perú y Colombia se enfrentaban diplomáticamente por la isla Santa Rosa, un video comenzó a circular en redes sociales con una promesa seductora: Alan García, el histórico líder aprista fallecido en 2019, parecía explicar desde una conferencia universitaria los fundamentos del reclamo peruano e incluso anticipar el conflicto en curso. El clip alcanzó 1.500 millones de visualizaciones en TikTok, más de 102.000 me gusta y 26.000 compartidos. Figuras como la periodista Martha Meier y el antropólogo Jaime de Althaus lo difundieron, dándole peso justo cuando la presidenta Dina Boluarte declaraba que la soberanía sobre Santa Rosa era innegociable.
El video era una falsificación. Los verificadores de La República identificaron la fuente original: una conferencia de historia de tres horas que García dictó el 1 de diciembre de 2003 en la Casa del Pueblo del APRA, disponible íntegramente en YouTube. Al comparar ambas versiones, quedó claro que los fragmentos citados correspondían a pasajes sobre la independencia de 1821 y la abolición de la esclavitud bajo Ramón Castilla. En ningún momento García habló de la isla Santa Rosa ni de Colombia.
Lo que había ocurrido era más sofisticado que un simple recorte: el audio había sido generado por inteligencia artificial entrenada con muestras de la voz de García para replicar sus características fonológicas. La organización juvenil Lengua Aprista lo confirmó en un comunicado público. La tecnología había sido lo suficientemente convincente como para engañar a millones de personas en plena crisis diplomática.
El momento no fue casual. El presidente colombiano Gustavo Petro había reabierto la disputa por Santa Rosa, y en ese clima de incertidumbre, el video fabricado llegó a ofrecer exactamente lo que muchos querían encontrar: la voz de un líder respetado que había previsto y explicado el conflicto. El caso ilustra una vulnerabilidad creciente: en un mundo donde la IA puede imitar voces con precisión creciente, incluso los muertos pueden ser reclutados al servicio de agendas políticas del presente, y la desinformación puede moldear la percepción pública antes de que la verdad tenga oportunidad de alcanzarla.
In mid-August, as tensions flared between Peru and Colombia over control of Santa Rosa island, a video began spreading across social media that seemed to offer historical vindication. The clip showed Alan García, the late former president and longtime APRA party leader, delivering a university lecture in which he appeared to explain Peru's claim to the island and even predict the current conflict with Colombia. The video accumulated staggering reach: 1.5 billion views on TikTok alone, along with over 102,000 likes and 26,000 shares. Prominent figures amplified it—journalist Martha Meier and anthropologist Jaime de Althaus among them—lending the false narrative credibility at a moment when Peru's president Dina Boluarte was asserting that Santa Rosa's sovereignty was non-negotiable.
But the video was a fabrication. Fact-checkers at La República traced the original source to a three-hour history lecture García delivered on December 1, 2003, at the APRA party's Casa del Pueblo. The speech exists in full on YouTube, recorded and preserved by party supporters. When investigators compared the viral clip to the original, they found the moments being cited—García discussing the creation of Peru's civil party after the purchase of freedom for Afro-Peruvians by Ramón Castilla, and the independence war of 1821. At no point in the actual lecture did García address Santa Rosa island or any conflict with Colombia.
What had happened was more insidious than simple editing. The audio overlaid on García's original video footage was synthetic—generated by artificial intelligence trained on samples of his voice to replicate his phonological characteristics. The APRA party's youth organization, Lengua Aprista, confirmed this in a public statement, identifying the manipulated audio as AI-created rather than authentic speech. The technology had been sophisticated enough to fool millions of viewers scrolling through social media during a live diplomatic crisis.
The timing was not accidental. Colombian President Gustavo Petro had raised the Santa Rosa dispute, prompting Boluarte's sharp response that Peru's sovereignty over the territory was beyond negotiation. In that heated moment, when citizens were seeking historical precedent and authoritative voices, the fabricated video arrived to provide exactly what people wanted to believe—that a respected former leader had foreseen and explained the very conflict now unfolding. The misinformation filled a vacuum of uncertainty.
The incident exposed a vulnerability that extends far beyond this single false claim. As artificial intelligence becomes more capable at mimicking human speech and appearance, the potential for coordinated disinformation campaigns during geopolitical crises grows sharply. A deepfake video can spread to hundreds of millions before fact-checkers can trace its origins and debunk it. By the time the truth emerges, the false narrative has already shaped public perception and potentially influenced diplomatic posturing. The García video demonstrated that in an age of AI-generated content, even the voices of the dead can be conscripted into service of present-day political agendas.
Citas Notables
The sovereignty of Peru is not up for discussion. The district of Santa Rosa de Loreto is and will remain Peruvian.— Peruvian President Dina Boluarte, responding to Colombian President Gustavo Petro
The audio was created using artificial intelligence that sampled García's phonological characteristics to emulate his voice.— Lengua Aprista, the APRA party's youth organization
La Conversación del Hearth Otra perspectiva de la historia
How did fact-checkers actually prove the audio was artificial and not just a different recording of García speaking?
They matched the viral video frame-by-frame to the original 2003 lecture on YouTube. The footage is identical—same room, same gestures, same moment in time. But the words being spoken don't appear anywhere in the three-hour original recording. The APRA party confirmed the audio was AI-synthesized by analyzing the voice characteristics.
So someone took old video of García and just dubbed over it with a fake voice?
Exactly. They sampled his actual voice from the 2003 lecture to train an AI model, then used that model to generate new speech that sounds like him. It's sophisticated enough that most viewers wouldn't notice the mismatch between his lip movements and the words.
Why would someone do this now, in August 2025, rather than earlier?
The Santa Rosa dispute with Colombia had just escalated. Petro raised it publicly, Boluarte responded sharply. There was real diplomatic tension and public anxiety. A video appearing to show García predicting exactly this conflict would seem like historical validation—proof that Peru's position was inevitable and wise.
Did the video actually change anything politically?
That's hard to measure. But it reached 1.5 billion people on TikTok alone. Even if most viewers didn't believe it, some did. During a border crisis, false historical narratives can harden positions and make negotiation harder.
What happens to the people who shared it?
The article doesn't say they faced consequences. Martha Meier and Jaime de Althaus amplified it, but there's no mention of them being held accountable. That's part of the problem—misinformation spreads faster than corrections.
Is this the first time someone's done this with García's voice?
The article doesn't indicate it. But it's likely not the last. Once you've proven the technique works and reaches that scale, others will try it with different political figures and different crises.