Mexico produces 65% of the dubbing in Latin America. And until now, no rule prevented an AI from copying the voice of its actors without paying or asking for permission. The government of Claudia Sheinbaum has presented this past February 13, 2026, an initiative to legally recognize the human voice as an artistic tool that cannot be cloned. If it prospers in CongressMexico would become something more than a government that looks after the interests of the actors: it would also be a world pioneer in regulation of voice cloning in a cultural setting.
Korea is to blame. The trigger for this reaction was not a native series, but some korean dramas. In May 2024, social media users shared fragments of Korean Prime Video series (‘My Boy is Cupid’, ‘The Beat of My Heart’ and ‘Field to Love’) denouncing an unusual feature: the dubbing into Latin Spanish sounded mechanical, robotic and without nuances. And there was also something very suspicious: there were no credits for voice actors anywhere. Without giving explanations, Amazon removed those dubbed versions and did not confirm the origin of the voices.
The straw that broke the camel’s back. It was a turning point: the voice actors guild had been denouncing for months how voice actors from all over the continent were losing jobs in favor of AI tools trained, in addition, with their own voices. Some actors, in fact, denounced the Kafkaesque situation that his voice was the one who had replaced him on a YouTube channel for which he worked.
Point of no return. In March 2025, Prime Video announced its AI dubbing pilot program in English and Latin Spanish. According to Amazonare twelve series that would not have been dubbed if it had not been for AI, presenting it as an opportunity for series to be seen that would otherwise remain unpublished. The suspicion of Latin professionals, as we have seen, went in a diametrically opposite direction. To calm things down, Amazon assured that localization professionals would monitor and correct the dubbed episodes with AI.
The protest. Mexico produces around 65% of the Latin Spanish dubbing destined for Latin America, according to data from the Mexican Association of Commercial Broadcasters (AMELOC), and has some thirty-five active studios with approximately 1,500 actors working. This human force was manifested last July in Mexico under the slogan “AI does not replace.” Among other requests, it was demanded that the voice be recognized as biometric data, at the level of a fingerprint. The purpose is to prevent its use without consent.
The proposal. According to the specialized media CO/AISince the summer of 2025, the National Copyright Institute (INDAUTOR) and the Legal Department of the Presidency have worked with more than 128 organizations to build a legal framework always in touch with the union. The resulting text reforms two existing laws: the Federal Labor Law incorporates dubbing actors and announcers as formal workers in the cultural sector, equating them to singers; and the Federal Copyright Law recognizes the human voice as a “unique and unrepeatable” artistic tool
That is, any use of it through AI requires express authorization from the owner, plus financial compensation. None of this prohibits dubbing with AI, it only protects the voices that train or replicate the model with mandatory contracts.
Missing. The initiative must pass the Chamber of Deputies and the Senate before becoming law, and it will take a while: the Mexican Congress accumulates proposals since 2020. There are more than sixty initiatives related to AI that have not yet received the corresponding legal response. Of course, this one seems to go faster: in November 2025, the Congress of Mexico City had already approved a similar opinionwhich reformed five federal laws.
Mexico, spearhead. This beginning of regulation in Mexico is an advance of what other countries are trying to regulate since 2023. For example, in 2024 in Tennessee, Governor Bill Lee signed the ELVIS Act to explicitly add voice among the attributes protected against unauthorized use with AI, something new in the US. The standard also holds responsible platforms that distribute tools whose main purpose is to generate voice replicas without authorization. California and New York have tried to regulate not the technology, but the contracts signed around these activities.
However, the limitations of these laws were soon demonstrated: in July last year, a New York judge did not rule in favor of two voice actors who discovered that their voices had been marketed as AI products. As it had not been made with a fixed recording, but with attributes such as tone, timbre or cadence, the court dismissed the claims. That ruling is the type of thing that the new Mexican legislation will try to avoid, and provide more robust protection to artists.
Header | Amin Asbaghipour in Unsplash
GIPHY App Key not set. Please check settings