Your Fear of AI Is Already Costing Us: What the Reform to Article 87 Reveals

View on LinkedIn

The Reform to Article 87 of the LFDA and the Lesson That Urgently Needs to Be Learned

Note: The reform to the Federal Copyright Law analyzed in this essay was approved by the Chamber of Deputies and by the Senate of the Republic, and is pending publication in the DOF. The analysis is prospective.

I.

Imagine a homeowners' association meeting called on an emergency basis on a Tuesday night. Twenty-eight residents in the lobby, some in their pajamas.

Mrs. Cristina — an eighty-two-year-old widow, a ground-floor resident since before any of those present moved in — had been burglarized the week before. Everyone loves her. Everyone agrees something must be done.

The residents of ground-floor units 1D and 1F, who live next door to Mrs. Cristina and arrived early with a printed proposal, take the floor. They suggest concentrating the entire security budget — cameras, a guard, sensors — on the ground floor. That's where the entrances are. That's where Mrs. Cristina lives. It makes sense.

The assembly approves by majority. It's late, everyone wants to go to sleep, and the proposal sounds reasonable.

Three months later, two residents run into trouble almost simultaneously. The tenant on the fourth floor has his car keyed in the parking garage. The tenant on the second floor, whose unit has a balcony facing the street, is burglarized. Both call the building management thinking the same thing: well, at least the cameras are there.

They are told the cameras covering their areas were disconnected in April. So what do we do? they ask.

They are told they voted to install new cameras only for the ground floor. The old system was deactivated.

This, translated into federal law, is what Mexico has just enacted.

II.

From the day the reform to Article 87 of the Federal Copyright Law is published, if someone uses your image without your consent for a campaign, a commercial, or a product, and you are not a performing artist or performer, you will have less legal protection than you had the day before.

Read that again. Less. Not more.

We need to stop here for a moment, because the story of how your image came to be protected by a copyright statute is a peculiarly Mexican one. In almost every other country, a person's image is regulated in the civil code, as a personality right. In Mexico, that protection was built on the Federal Copyright Law — not because it fit perfectly there, but because that was where a functioning legal regime already existed. For decades, Article 87 was the lock. Jurisprudential thesis 166/2023, arising from the Arjona case, extended it to everyone: any person, not just celebrities, could invoke Article 231, Section II against anyone who used their face or any element of their identity without consent.

The reform unlatched that lock and installed a new, reinforced one — but only at the entrance. Only for the ground floor.

And that change carries far too many problems. Starting with the most obvious: what exactly is a performing artist or performer? Who decides? Do they need to have released an album? A union card? A minimum number of performances per year?

What happens to public figures, influencers, athletes, or simply ordinary people?

For all of them, the new law's answer is: go to the civil code, where personality rights are regulated — or where they should be regulated. But here's the surprise: there is nothing there. The system was built, perhaps by accident, on Article 87. And only Mexico City has a specific statute. In the rest of the country, the work starts from scratch.

How is this an improvement over what we already had?

How we got here is, precisely, the story of the homeowners' association meeting.

III.

Mrs. Cristina of the ground floor is called, in the federal context, Pepe Lavat.

On June 2, 2025, Mexico voted for its judges. The following day, the INE uploaded a video to TikTok thanking those who had participated. Over the footage, an unmistakable voice: deep, measured, with the cadence of an epic narrator. The voice that for decades told Latin America that Goku was preparing for a new battle. https://www.youtube.com/watch?v=jRuQnORxJqc

Except Pepe Lavat had died in 2018.

His widow, Gisela Casillas — herself a dubbing actress — asked for an explanation. The INE's response, as she recounted it, was terse: then sue us.

Five weeks later, more than a thousand dubbing actors and voice-over artists marched at the Monument to the Revolution. They carried signs reading "AI Doesn't Feed Itself" and "One Voice." The residents of 1D and 1F had arrived at the assembly. The president listened to them. Five months later, the Executive submitted the bill to reform the LFDA. In April 2026, Congress approved it.

The dubbing industry got what it asked for. And then some. At the expense of everyone else.

IV.

There is an uncomfortable detail that almost no one mentioned.

The INE video was never a commercial use. It was the impulse of whoever was running the Institute's social media accounts, trying to be clever, triggering an avalanche that could no longer be stopped. But Article 87 protects against unauthorized commercial uses of a person's image and voice. A government agency producing an institutional piece did not fit — and it did not fit before the reform, and it will not fit after it either.

This had happened before. In 2014, COFEPRIS included a photo of Soledad Félix — taken during her recovery from a heart attack caused by smoking — on cigarette packages, as part of the health warnings. No one asked her permission. She appeared at one of the worst moments of her life. Soledad filed a complaint, sought compensation with the support of the CNDH, and the case exposed exactly the same gap: COFEPRIS was not profiting, it was not advertising, there were no commercial purposes. It was a public policy that used a person's image without her consent.

Article 87 did not cover that case. Nor did it cover the INE case. The actual legal gap — the one the dubbing industry believed it was closing with this reform — was not in the Federal Copyright Law. It was in civil law, where personality rights are regulated in nearly every country and where Mexico never finished building the regime.

And there is something even more absurd. In both cases — COFEPRIS and the INE — the infringer was the government itself. The same authorities that should safeguard the most basic rights, not even as a minimal matter of personal data protection, are the ones that got us into these predicaments. It was not a private company profiting. It was the State using Soledad Félix's face without permission. It was the State using Pepe Lavat's voice without permission. And then it is that same State that approves a reform that still does not resolve what it itself caused.

It is as if the person who robbed Mrs. Cristina worked for building management. And then, in response, management called the assembly to overhaul the camera system.

Something led the dubbing industry to believe it was unprotected simply because the word "voice" did not appear expressly in Article 87. But the interpretation built by jurisprudential thesis 166/2023, arising from the Arjona case, already went further than that. And even so, that interpretation would not have applied to the case that mobilized them. The Lavat case was not an Article 87 case.

The reform did not solve the problem it claimed to want to solve. And it unlatched the lock that was working for everyone else.

Returning to the building: Mrs. Cristina did not need the entire budget redirected to her. She needed two additional cameras in her hallway, a new deadbolt, better lighting. She needed more security, not the same security concentrated on her at everyone else's expense. And least of all a new system that did not even aim at the actual point of entry.

V.

George Stigler described in 1971 a well-known pattern in regulatory theory: when an industry is organized and the public is dispersed, the industry captures regulation. Laws come out drafted in favor of those who had a seat at the table.

What happened with Article 87 is that pattern. The residents of 1D and 1F arrived at the assembly with their printed sheet. They had reasons. They had grievances. They had Mrs. Cristina as their banner. The other residents barely understood what the meeting was about.

Article 87, before the reform, spoke of a person's likeness. Any person. It was universal. It covered every floor.

The reformed Article 87 reads: the image, including the voice, of performing artists and performers, as well as their characters.

Read that again. Performing artists and performers.

The cameras that covered the entire building now cover only the ground floor.

No one in the assembly wanted this. The residents of 1D and 1F are not villains. Mrs. Cristina asked for nothing. The others raised their hands in good faith. And still, the outcome is unjust.

That is the tragedy.

VI.

Upon reading this, someone might say:

Ricardo is against the poor dubbing artists and the artistic industry.

My response, direct and open, is yes. I am.

I have enormous respect for their craft and their work. But the way they pushed this reform through did not merely privilege them and create tiers of citizenship — it dragged everyone onto the dance floor. And they did it over something that did not need to be over-regulated, with deplorable legislative draftsmanship, out of fear of AI. They left everyone else outside.

So yes: I am opposed. But not to the industry, not to their work, not to their right to defend themselves against AI. I am opposed to the means.

I am opposed to a reform that uses a legitimate cause to build unequal protection. I am opposed to a legal regime that depends on who arrived first at the assembly.

What I do not want is for legal classes to be created by trade or industry.

The question was never whether to legislate. The question is how.

And here the Article 87 case ceases to be a technical matter of intellectual property and becomes a case study in how not to legislate artificial intelligence. Because many more assemblies are coming: AI will affect translators, illustrators, musicians, photographers, lawyers, physicians. Every industry will arrive at the building asking for the budget to be directed to its hallway.

If the Article 87 model is the template, we will end up with a building where security depends on the resident's occupation. Every profession with lobbying capacity will secure its protected floor; the rest of us will be left without cameras.

This contradicts something elemental about law in a market economy: rules must be, as far as possible, universal. Regulatory fragmentation distorts competition, generates rents for protected industries, and delays adaptation. The Luddites smashed looms in 1811 and lost anyway. Legal protection against technology rarely saves an industry; what saves it, when anything does, is adaptation.

Mexican dubbing actors have a genuine competitive advantage — quality, tradition, neutral Spanish — that no AI will replicate in the near term. That advantage is better defended with sophisticated contracts, strong unions, and a universal regime for human likeness, than with exclusive cameras for one floor.

VII.

There is a better way. The U.S. Copyright Office published it in July 2024: regulate conduct, not technologies; recognize a universal right of digital reply with federal protection; quick remedies rather than punitive offenses.

Mexico could have done something like that. It could have bought more cameras. Covered every floor. Created copyright-style notice and takedown procedures, expedited arbitration, specific injunctive relief for deepfakes, transparency obligations for platforms.

It did not. It did what those who arrived at the assembly asked for. It confused protecting with privileging.

And it left a methodological precedent. The next assembly — on AI and copyrighted works, on AI in research, on AI and personal data — will look just like this one if no one corrects the method.

VIII.

Back to the building.

Three months after the assembly, a couple of residents run into trouble. They call the building manager. They are told the old camera system was deactivated when they voted to install new ones only for the ground floor.

That is what happens when we legislate out of fear. Mrs. Cristina ends up better protected. So do the residents of 1D and 1F. The rest of the building pays the same dues and is left without cameras.

AI is going to force us to make many legal decisions in the years ahead. The question is not whether we will regulate — we will — but how. If we do it assembly by assembly, industry by industry, we will end up with a building where security depends on occupation. If we do it with universal principles, we might build something different: a regime that recognizes image and identity as human attributes before they are professional assets. Cameras in every hallway.

The reform to Article 87 is the first major assembly of the AI era in Mexico. Now is a good time to ask ourselves whether we want the rest of the building governed by its rules.

Meanwhile, somewhere on a server, an AI is being trained on the photos of someone who lives on the second floor with a balcony facing the street, or on a floor whose parking spot is in the basement. He is not famous. He will never attend an assembly. And one day his face will appear in an advertisement he never authorized, and he will call the building manager to ask what he can do.

Let us hope the answer is not, once again, that the old system was deactivated.