Rebuilding faces and identifying tattoos, AI joins the search for the missing in Mexico
By Karen Esquivel, CNN en Español
(CNN) — In 2021, Héctor Daniel Flores Hernández was reported missing. Two years later, his father, Héctor Flores, saw his son’s image speak again. In a video generated through artificial intelligence technology, which digitally animated a still image and gave it a synthesized voice, the figure of Héctor Daniel Flores Hernández narrated the story of his disappearance — when he was 19 years old from his home in Guadalajara — and demanded that the authorities find him alive. Since then, his father has made the search for a way of life.
“I couldn’t finish watching it the first time,” Héctor Flores said in an interview when the video was first produced. “It took me a while to process it. Seeing them with the last photo you have, saying what you know from investigations, is heartbreaking.”
The video was part of an initiative launched in 2023 by the collectives Luz de Esperanza and Alas de Libertad — groups of relatives of missing persons who organize to carry out searches, raise awareness of cases and demand justice — in the western state of Jalisco, to give a voice to the missing-persons notices about their relatives. Flores, a co-founder of Luz de Esperanza, said that the initiative continues and that “it is a perfect tool not only for the search but also to raise awareness and try to create empathy.”
Disappearances are all too common in Mexico, a country that has recorded more than 132,000 missing people since the National Registry of Disappeared and Unlocated Persons of the Secretariat of the Interior began keeping track in 1964. Human Rights Watch reports that the government has not taken sufficient measures to prevent disappearances or punish those responsible.
President Claudia Sheinbaum has acknowledged that most disappearances are linked to organized crime, and Amnesty International notes that the problem can be attributed to overall violence and insecurity. There is no complete official data set on the specific causes of all disappearances; the national registry includes data on past disappearances committed by the authorities against leftist groups and guerrillas, but the figures are mostly from recent years, when the fight against criminal organizations intensified.
In March 2025, Sheinbaum’s government announced an array of new initiatives to respond more quickly to disappearances, to treat disappearances as seriously as known kidnappings, to make statistics more available and to improve assistance to victims.
The video initiative, meanwhile, is part of a new generation of projects that have turned to AI or machine learning for help with the crisis. Universities, search collectives, other organizations and government authorities have developed and implemented AI to investigate missing persons, using techniques including database analysis, forensic identification or age-progression projections.
“The goal is that these tools be useful for entities that make up the national search system, such as prosecutors’ offices, commissions, and the Semefo (Forensic Medical Services), to assist and facilitate the work of the people,” said Andrea Horcasitas, head of the Human Rights Program at Universidad Iberoamericana in Mexico City. The program is part of the Consortium for the Ethical Use of AI in the Search for Disappeared Persons, which was created in October 2025 with the aim of discussing the use of techniques for searching missing people in Mexico and reflecting on the responsible use of AI in this area.
AI to identify tattoos
The Public Policy Collaborative Solutions Laboratory (Lab-Co), a non-governmental organization based in Mexico, works on finding new solutions to problems of security, violence and access to justice in Latin America. The laboratory has developed three projects that incorporate artificial intelligence in the search for disappeared persons.
The first is IdentIA, a tool that uses AI to identify and classify photographs of tattoos on unidentified bodies, explained Thomas Favennec, the executive director of Lab-Co.
Users can search the files with text, using the words in a tattoo or a verbal description of one, or through image-based searches that compare photographs directly against a historical database of tattoos belonging to unidentified people.
“It doesn’t matter the quality of the tattoo or the angle. This system works using a vector search and does not require the internet; no sensitive content is uploaded to the internet, which ensures that no one’s information is compromised,” Ángel Serrano, Lab-Co’s data and technology coordinator, said.
Serrano demonstrated how IdentIA conducts a tattoo search in a matter of seconds and enables cross-referencing with reports of missing persons.
This system was incorporated this week into missing-persons database systems in the state of Jalisco and is in the process of being implemented in the forensic services of Quintana Roo and Zacatecas.
Searching across separate files
Another of the lab’s tools, Favennec said, is ContextIA, which makes it possible to process multiple unstructured documents from investigation files and extract clear data. ContextIA is capable of answering specific queries with direct quotes and page references; extracting particular data such as phone numbers, license plates and coordinates; and, as its name suggests, cross-referencing cases in the context of different databases.
The third tool allows for the analysis of names in a structured and more powerful manner, allowing users to find matches in different databases where names may be written in different ways. Favennec said that, to date, it has been applied to identified deceased persons who have not been claimed.
“Families don’t know this happens, and it is complicated because there is an issue with databases that do not intersect,” he said. “There’s a challenge in finding the family. So what we have developed is something that allows for these comparative analyses between the situation of missing persons and the records of forensic services in different states across the country.”
The initiatives are part of a project called “Building Alliances for the Search for Disappeared Persons and Human Identification,” funded by the European Union and the British Embassy through the Frontier Tech Hub initiative.
Restoring faces of the disappeared
Horcasitas, from the Universidad Iberoamericana, points out that some prosecutors’ offices are already using a tool called ImageBox, which cleans up the images of the faces of people found in the morgue or the Forensic Medical Service and the Institute of Expert and Forensic Sciences of Mexico City, to aid in the identification of individuals.
“This prevents people from being forced to look at heartbreaking photographs, which obviously have psychosocial impacts on anyone who has to go through morgue catalogs,” she said.
The Mexico City Prosecutor’s Office has used the “inpainting” technique, which uses artificial intelligence to fill in, restore or remove selected parts of an image, allowing damaged areas to be added or repaired with realistic results. With this, authorities carry out reverse searches, issuing a bulletin to find the person’s family.
CNN contacted the capital’s Prosecutor’s Office for more details, but so far has not received a response.
Children’s faces grow up
The Regresa Project, developed by researchers from the National Autonomous University of Mexico (UNAM), seeks to generate an image of how a child who disappeared years ago would look today, to help their families find them.
When a person disappears, the standard search protocol is based on the most recent photo available. Children and adolescents who are not quickly located may not be recognized in the medium and long term, due to the speed at which their faces and bodies change and mature.
Led by Ana Itzel Juárez Martín, PhD in anthropology from the National Autonomous University of Mexico (UNAM), the initiative, which is still a pilot program, combines AI tools with knowledge and techniques from physical and social anthropology.
The program’s algorithm is designed to carry out age progression using photos of missing children to determine what they would look like at age 5, 15 or 30. It could also be used to show what a current adult looked like in childhood.
Official figures count more than 118,000 minors aged 17 and under reported missing between 1964 and September 2025. A 2022 report from the Network for Children’s Rights in Mexico points out that in the case of missing minors, recruitment networks may integrate them into organized crime or draw them into sexual exploitation and trafficking.
Juárez Martín explained that the goal of the Regresa Project is to train the algorithm to learn the natural biological growth of the face, specifically guided by facial shapes and growth patterns commonly found among people in Mexico.
“Although there are already some image banks for research use, there are none solely of the Mexican population,” she said. “So this algorithm would be the first of its kind and would be trained to identify the variability that exists among Mexicans, for example, the type of nose, thickness of lips, shape of eyes and eyebrows.”
Cooperating and working faster
In a country where, according to the organization Causa en Común, 41 people disappear every day, it is important to have tools that assist and facilitate the search for missing persons throughout the country.
What’s next is to keep looking for gaps or holes where certain processes need to be made more efficient. “We are waiting to see how these artificial intelligence tools operate, to see if they work, the improvements that need to be made, and also to understand what the workflow is within the organizations that make up the National Search System,” said Horcasitas.
Favennec pointed out that the implementation of AI in search and location processes has been well received by groups and authorities, but added that it is necessary to know that the technology “is not magic, it helps to process information faster and better…In a crisis where so many things are mixed in so many different ways, what is needed is collaboration.”
The-CNN-Wire
™ & © 2026 Cable News Network, Inc., a Warner Bros. Discovery Company. All rights reserved.
