From Sheinbaum to Meloni: world leaders push measures to protect themselves from AI deepfakes

Heads of state and politicians worldwide are advancing measures to protect themselves from deepfakes created with artificial intelligence. Some are pushing regulations that cover all officials in their countries. Others are simply taking steps to prevent their voices from being cloned.

These provisions are being rolled out as it emerged that at least 46 female members of the European Parliament, one in six, have appeared in hyperrealistic, fake pornographic images generated with artificial intelligence, according to an investigation by the organization Agora Digitale Transformation presented Monday at the European Parliament.

“Violence against women, online or offline, is intolerable and deplorable,” said European Parliament President Roberta Metsola, urging member states to ensure effective enforcement of new EU rules, to act against apps that allow the creation of this kind of content and to prosecute the distribution of nonconsensual sexual deepfakes.

In Italy, Prime Minister Giorgia Meloni has applied to register her voice as a trademark with the European Union to defend herself against AI-manipulated montages. According to an advance report Monday by the newspaper La Repubblica, the far-right leader filed a recording of four seconds with the European Union Intellectual Property Office (EUIPO) documenting her voice and its inflections, in which she says the phrase Io sono Giorgia Meloni (I am Giorgia Meloni) twice. The application is still under review, but if accepted, Meloni would gain new legal tools to block the spread of false content, such as demanding the removal of illicit material and later claiming compensation for damages.

Elena Gil, an AI lawyer and founder of Data Guardians, warns that registering a voice does not provide absolute protection against the generation of fake videos: “It is true that the trademark can help in certain commercial uses, but because protection is granted for the products or services for which you have registered it, it does not cover every deepfake,” she explained.

While Meloni has chosen an individual protection tool, in Mexico the government is working on a much broader measure that would protect more than just President Claudia Sheinbaum. The Senate last Wednesday approved, by 72 votes in favor and 34 against, a reform to penalize the use of fake videos depicting government officials. The amendment, which still must be approved by the Chamber of Deputies, has been dubbed the “anti-meme law” and has drawn harsh criticism from the opposition. Lawmakers accused Sheinbaum’s administration of creating an instrument to censor satire and parody by reforming the Federal Law on Protection of Industrial Property and the federal Penal Code.

Faced with the uproar, Sheinbaum herself tried to contain the controversy. “What is being sanctioned? Using the image of the president in a video using artificial intelligence to commit a fraud. That’s what is in the law: that crimes cannot be committed. Not political satire, not even personal insults,” she said at her Thursday morning press conference. The federal government’s legal adviser, Luisa María Alcalde Luján, said the approved text provides for a prison sentence of five years and a fine of up to 1,173,000 Mexican pesos (about €58,000).

Other governments’ responses to protect their leaders and citizens from the dangers of artificial intelligence vary by country. Elena Gil, a communicator and AI lawyer, highlights Denmark’s case. “They have passed a law that gives citizens rights over their own digital identity — voice, face or body. This is to prevent those elements from being used to train AI. The regulation even allows them to seek compensation. As an exception, satire or parody using AI-created photos or videos is permitted, because not every generated image necessarily crosses the threshold into illegality,” she noted.

In Brazil, measures have focused on the electoral sphere. In August, the Superior Electoral Court (TSE) reached an agreement with the British company ElevenLabs to create a protection mechanism to block unauthorized cloning of candidates’ voices. Key aspects of the partnership include the fight against disinformation, fraud detection and cooperation in judicial investigations to determine whether suspicious audio recordings were generated using the company’s own systems.

And in Spain, digital law expert Borja Adsuara feels that deepfakes should not be regulated as a separate category, as the government of Pedro Sánchez had intended to do before dissolving parliament and calling a snap election. “A new right to one’s image in the digital environment should not be created. The Spanish Constitution already protects us in its article 18, as well as civil protection laws, the penal code and data protection laws. Image and voice are personal data,” he insisted in a telephone interview, noting that the Spanish Data Protection Agency has already opened proceedings for use of a person’s image and voice without their consent.

“Instead of using the term deepfake, you can use realistic or real images. With that terminology, it does not matter whether they were created with artificial intelligence or with Photoshop. You should not have to reform the law every time we face an emerging technology,” he notes.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition