Artificial intelligence (AI) has become a new resource for high-risk criminal groups in Latin America. A study published in September 2025 by researcher Juan Manuel Aguilar Antonio of CISAN–UNAM, as part of the European PACCTO 2.0 program, documents that organizations such as the Sinaloa Cartel and the Jalisco New Generation Cartel (CJNG) use algorithms, bots, and automation systems in their operations, from extortion to drug trafficking logistics. The analysis was published in the official Zenodo repository of the European Commission and CERN.
In contrast, another academic article from the National Institute of Criminal Sciences (INACIPE) in Mexico, entitled “Artificial Intelligence for the Detection and Prevention of Human Trafficking,” proposes that the State could use AI to prevent complex crimes such as human trafficking through early detection algorithms. Both documents expose a dilemma: while cartels are already capitalizing on the technology, institutions are still trying to define how to apply it legally and ethically.
High-Risk Criminal Networks and Their Leap into AI
The PACCTO 2.0 study, titled Use of Artificial Intelligence by High-Risk Criminal Networks, identifies Mexican cartels as pioneers in the use of AI in organized crime contexts. According to Aguilar, these organizations have begun to integrate internal structures dedicated to the use of advanced technology, such as the CJNG’s drone unit, which combines AI for surveillance and improvised weapons.
The research classifies criminal groups into four types based on their technological adoption:
Classic hierarchical groups, which incorporate AI to improve logistics.
Cyber-collective groups, which operate primarily in digital environments.
Autonomous platforms, which sell AI-based criminal services under the Crime-as-a-Service model.
Parastatal actors, which employ AI for territorial and political control.
The CJNG and the Sinaloa Cartel fit into the first category, but are already showing trends toward the others.
Types of Use Detected
The study documents specific applications of AI in criminal operations:
Automated extortion: Bots and voice cloning allow for simulating virtual kidnappings and pressuring victims without the need for physical contact.
Deepfakes and digital manipulation: Generation of fake audio and video content to deceive or discredit adversaries and authorities.
Illicit route optimization: Algorithms that model drug transport routes while avoiding border controls.
Digital fraud and smishing: Use of AI to create personalized phishing campaigns via SMS or email.
Hacking and Adversarial Intelligence: Documented cases include access to video surveillance systems in Mexico City and data theft from an FBI agent’s phone, attributed to the Sinaloa Cartel, according to the U.S. Department of Justice.
The document warns that these capabilities allow cartels to reduce costs and risks by shifting part of their operations to the digital realm.
AI-Based Criminal Platforms
The study also mentions the growth of clandestine AI platforms designed for criminal use, such as FraudGPT, DarkBARD, and Xanthorox AI, which offer identity cloning, malware generation, and data manipulation services. These systems are sold under the Crime-as-a-Service model, making it easier for criminal groups to access technology without having to develop it.
The risk, Aguilar warns, is that these platforms “normalize” the availability of criminal AI, increasing the capacity of local and global actors to carry out sophisticated crimes.
The Contrast: AI to Prevent Crimes
While the PACCTO study documents how cartels capitalize on AI, the article published by INACIPE in the Mexican Journal of Criminal Sciences (2025) proposes the use of the same technology to prevent human trafficking.
The text suggests that predictive algorithms could detect patterns in fake job advertisements, suspicious interactions on social media, or financial transfers linked to exploitation. The goal would be to anticipate risks and enable early interventions.
INACIPE proposes complementary measures:
National data-sharing protocols.
Training prosecutors in digital analysis.
Collaboration with technology platforms.
Ethical oversight to avoid bias in algorithms.
Ethical and Legal Risks
Both documents agree that the use of AI, by both criminals and authorities, poses ethical risks. In the field of organized crime, the technology can increase the capacity for intimidation and manipulation. At the state level, there is the possibility of biases that criminalize victims or violate citizens’ privacy.
Therefore, Aguilar recommends establishing regulatory frameworks that limit criminal access to these tools, while developing public detection and defense capabilities.
Preventive Measures Against Criminal Use
The PACCTO 2.0 study suggests several actions that States could implement:
Digital forensic units specialized in AI to track online criminal activity.
Public-private cooperation with technology and telecommunications companies.
International audits of algorithms to detect malicious platforms such as FraudGPT.
Cybersecurity and digital intelligence training programs for law enforcement agencies.
Regional cooperation strategies in Latin America and Europe to track data flows and transnational operations.
Global and regional scenario
The report highlights that Latin America has become fertile ground for the criminal adoption of AI due to regulatory gaps and technological inequalities. At the same time, the region faces the urgency of incorporating AI into state strategies to prevent cartels from maintaining their advantage.
In Europe and North America, there are already pilot projects that use AI to detect human trafficking and digital fraud. Mexico, however, still lacks a consolidated framework that combines technological innovation with legal oversight.
A dispute over artificial intelligence
The findings of both studies reflect that AI has become a contested field. For organized crime, it represents a power multiplier that automates crimes and expands its global reach. For the State, it offers the possibility of anticipating crimes and protecting vulnerable sectors.
The difference, for now, lies in speed: while cartels are already applying AI in a practical way, institutions are barely building the protocols to do so.
The challenge, the authors conclude, will be to ensure that AI is not captured by criminal organizations, but rather becomes a regulated, supervised resource geared toward protecting society.
Source: proceso
