A bundled bill, four distinct policies
On April 15, 2026, the Mexican Senate approved the new Federal Law of Cinema and the Audiovisual by 87 votes to 1 with 18 abstentions, abrogating the 1992 Federal Cinematography Law and turning the bill over to the executive for promulgation. Champions and critics on the Senate floor agreed on one thing: the bill bundles four discrete policy levers into a single statute, each of which deserves a separate verdict.
Those levers are:
- A 10% screen-time quota for Mexican works in theaters, with equitable schedules (no banishment to graveyard slots);
- A minimum 14-day exhibition window for domestic releases, doubled from the prior seven days;
- A mandatory, easily locatable Mexican-productions section in streaming catalogs — directly affecting Netflix, Disney+, Amazon Prime Video, HBO Max and ViX;
- Article 29, which prohibits AI-generated dubbing of foreign cinematographic and audiovisual works, requiring dubs to be performed by human "artistas intérpretes ejecutantes."
The strongest case for the law
The most defensible argument starts with discoverability, not coercion. Streaming algorithms optimize for global engagement, and Mexican productions often sit several scrolls deep in catalogs sized at tens of thousands of titles. A clearly labeled, default-visible "Mexican works" rail is a thumbnail-level nudge, not a content restriction. It does not tell platforms what to produce or buy — only how to surface what they have already licensed.
The dubbing-industry case is also real. According to the National Association of Actors of Mexico, the country employs roughly 10,000 dubbing artists across 43 studios — a regional voice-work hub that supplies the bulk of Latin American Spanish dubs. AI synthesis can reproduce a recognizable voice for a few cents per minute, where a human session runs $50–$300 per minute. Without consent-and-compensation rules, voice talent risks being commoditized without recourse — a labor concern any pro-worker regime should take seriously, and the immediate cause that earlier passed Article 29 through the Chamber of Deputies in late March.
Where the law overshoots
The first overreach is conceptual. Article 29 does not regulate AI dubbing — it bans it. The right benchmark for regulation is the harm (non-consensual voice cloning, uncompensated likeness reuse), not the input modality. A consent-and-compensation regime — analogous to the AI rider from the 2023 SAG-AFTRA strike — could have protected dubbing artists from misuse without prohibiting an entire technological category. Forbidding AI dubbing wholesale forecloses use cases that don't harm anyone: accessibility variants for visually impaired viewers, late-night turnaround of news content, dubs of works whose original cast is dead, and synthetic voicing performed under license with the original human artist's consent and royalty share.
Second, the screen-time and 14-day rules port a 20th-century cultural-protection instrument into a market where the binding constraint is no longer theater shelf space. Doubling the theatrical hold assumes underperforming domestic releases benefit from forced retention. In practice, comparable European hold-back rules have mostly transferred revenue between titles rather than from foreign to domestic releases. Latin American audiences have been migrating to streaming for years; the relevant cultural battleground is the home recommendation grid, not the multiplex schedule.
Third, the OTT visibility rule is broadly sensible but vague. It tells platforms to create a Mexican-content rail without benchmarks for prominence, without clarity on whether the rail must be defaulted-on, and without carve-outs for platforms that already exceed reasonable thresholds. The Ministry of Culture is empowered to conduct weekly and semi-annual reviews, fine on paper, but absent published metrics it risks devolving into discretionary jawboning. Even floor speakers from PAN — who voted to abstain rather than oppose — noted that platform oversight is "insufficient" and that the bill says nothing concrete about AI in production, only in dubbing.
A predictable cost transfer
The AI-dubbing prohibition will not preserve voice-actor employment at the volumes the law's authors hope. The likely industry response is to dub fewer foreign titles into Latin American Spanish, delay regional releases, or default to subtitles for catalog content no longer cost-competitive to localize. The Consumer Choice Center, opposing an earlier 15% national-content proposal in 2021, warned that a strict threshold could force Prime Video to "delete two-thirds of its library" in Mexico. The new 10% rail is gentler, but the directional incentive is identical: less content choice, higher compliance overhead, and prices that will rise to absorb both.
There is a comparative warning. The EU's AVMSD requires at least 30% European works in streamer catalogs, and Canada's Online Streaming Act added discoverability rules under the CRTC. Both have generated multi-year regulatory disputes, opaque enforcement metrics, and reduced platform investment at the margin. Mexico, with a smaller home market than either, will feel those frictions more acutely.
A proportionate version
A proportionate version of this law would keep three things and drop one:
- Keep the visibility rail for Mexican productions, with measurable, published prominence benchmarks tied to a sunset review;
- Keep the consent-and-compensation regime for voice-actor likenesses, drafted as a tort-style right rather than a technology ban;
- Keep the production tax credit (separately enacted in February 2026 by DOF decree) — supply-side incentives travel better than demand-side mandates;
- Drop Article 29's categorical AI prohibition in favor of a non-consensual-cloning rule.
The cleanest path forward is implementing regulation that reads Article 29 narrowly — as a non-consensual-cloning rule, not a technology prohibition — and that publishes concrete discoverability metrics before the screen-time obligations attach.
The bill now heads to the executive for promulgation in the Diario Oficial. The good policy and the bad arrived bundled. They will be litigated bundled too.