Mexico needs an intelligent regulation of artificial intelligence. One that protects without blocking, that sets limits without inhibiting, and that understands that the problem is not in how the technology is trained, but in how it is used. Regulating the input is to close the door to the development and use of Artificial Intelligence. The Executive's initiative to reform the Federal Labor Law and the Federal Copyright Law starts from a legitimate objective: to protect the image and voice of people in the face of the use of artificial intelligence. The risk we face is clear: in an attempt to protect rights, we could end up suffocating innovation. However, the problem lies in how it intends to do so. The proposal establishes that any use of image or voice—including its use in artificial intelligence systems “known or to be known”—will require express authorization. But it is in the output—the results—where the relevant damages occur: identity theft, undue commercial exploitation, or deceptive use of voice and image. Regulating training is intervening in an abstract phase, difficult to delimit and practically impossible to control without generating serious side effects: uncertainty, litigiousness, and disincentives to innovation. Instead, regulating the results allows for the legal intervention to be focused where it really matters: when there is a concrete damage, an identifiable victim, and a sanctionable conduct. In that case, the responsibility does not lie with the technological tool or with the one who trained it with data of various kinds. It is the person who makes a bad use of the technology—for example, using a person's image or voice to slander them—who must be sanctioned. Furthermore, the initiative introduces dangerous ambiguities—such as the scope of the concept of 'publication'—and may transfer responsibilities to developers for the use that third parties make of their tools, which even strains international commitments such as those derived from USMCA. This data is not minor, especially in a context in which the regulatory trend in the United States is going in the opposite direction. The problem is conceptual. This formulation, far from offering certainty, introduces an expansive regulation that could end up inhibiting technological development. It is no coincidence that the American Chamber of Commerce of Mexico (AmCham México), the Latin American Association of Internet (ALAI) and the National Chamber of the Electronic, Telecommunications and Information Technologies Industry (CANIETI) have warned that such a broad wording generates legal uncertainty in dynamic digital environments and can affect innovation. I suppose that large technology companies and platforms such as Meta or Google will have the same concern. Regulating the output is doing it where it really matters. P.S. Yesterday at the Faculty of Law of the UABC in Tijuana, the first presentation of my book Totalitarian Total took place.
