Mexico

It remains unclear whether any of these data-monitoring systems are used for targeted surveillance rather than mass surveillance. Although state authorities claim that the cameras do not store biometric data on everyone—except for individuals reported missing or suspected of criminal activity—some critics argue that, in practice, broader biometric collection and retention occurs. These concerns echo earlier allegations that Mexican authorities unlawfully surveilled ordinary citizens using spyware such as Pegasus or Titan, including claims that private data and geolocation information were leaked or sold to criminal organisations.

In addition to these state-level efforts, the federal government has taken steps to lay the institutional and legal groundwork for deeper interconnection between state and federal systems — broadening the datasets from which AI tools would draw when analysing and anticipating criminal activity. The Law on the National System of Investigation and Intelligence in Public Security, which entered into force in July 2025, is the central piece of this new architecture. One of its core purposes is to establish a Central Intelligence Platform — operated by the National Intelligence Centre — that interconnects all of the Mexican State's security-related databases, including vehicle registries, biometric records, tax and banking information, telephone data, weapons registries, health records, and property data. In certain cases, data held by private parties may also be incorporated. This aggregated information is then to be processed and analysed using AI tools to generate intelligence products and operational strategies capable of supporting special operations and supplying the Attorney General's Office with evidentiary inputs for criminal proceedings at both the federal and state level.

Prosecutors

Mexican prosecutors have also begun to explore more targeted uses of AI, from proposed institutional reforms to pilot tools supporting case management and evidence review.

An unlegislated proposal, introduced by opposition legislators in 2023, would create a Specialised Prosecutor's Office for Cybercrime and Artificial Intelligence within the Attorney General's Office responsible for investigating and prosecuting federal cybercrimes, including offences committed using AI. The bill was referred to the Justice Committee but was not subsequently approved.

Case management

As at July 2026, there is no official implementation of AI-driven triage or automated case-prioritisation on a national scale. However, certain systems have been trialled regionally. In 2024, for example, the Office of the Attorney General of the State of Zacatecas, working with researchers from the Tecnológico de Monterrey, deployed a machine learning tool to help prosecutors manage chronic case backlogs. Fed by administrative data from the office's digital case management system, the tool produces two weekly outputs: a ranked list of 300 open cases predicted to be resolvable within six months and a complementary list of the 1,000 cases least likely to progress, cross-referenced against statutory prescription deadlines. Prosecutors retain full discretion over how to act on either list, but are expected to give priority to a larger portion of easy cases as a means to reducing the pressing backlogs. In retrospective validation, the model correctly identified approximately 74% of the cases it flagged as likely to resolve.

Evidence review and analysis

In March 2026, the Federal Attorney General's Office launched a tool known as the National Mapping System (Mapeo Nacional), designed to support prosecutors with contextual information on crime. The system relies on AI to process large volumes of data drawn from investigation files, institutional platforms, and open-source intelligence gathered from social media networks and international migration databases. By cross-referencing these inputs, the tool detects connections between crimes that previously appeared unrelated, mapping high-incidence hotspots, and identifying behavioural patterns among criminal networks. The system is primarily oriented toward investigations involving enforced disappearances, crimes against migrants (including trafficking, kidnapping, and extortion), torture, violations of freedom of expression, and the location of clandestine detention centres.

The Financial Intelligence Unit (Unidad de Inteligencia Financiera, UIF) and the Attorney General's Office have formally classified information related to their AI use; the Attorney General's Office has stated that disclosing technical details could enable adversaries to manipulate its models. These classifications came to light through a review of forty-five freedom-of-information requests submitted to federal agencies, which found that at least fourteen government institutions were using or developing AI as of November 2025.

Courts

Mexican courts are increasingly adopting AI. This adoption follows digitisation efforts of the Mexican judiciary. A consolidated digital case-management infrastructure operates within the Federal Judiciary (Poder Judicial de la Federación) (PJF). This includes systems that enable electronic filing, electronic case files, electronic notifications, and remote procedural hearings. These services operate under the judiciary’s ‘Justicia Digital’ framework and its online services portal. Functionally, these systems facilitate document submission, service of process, case tracking, and digital access to court records. These tools facilitate electronic workflow and service-delivery platforms.

Case management

Several Mexican courts have developed their own AI tools for administrative or support functions related to case management. For example, the State of Mexico has developed an automated system that anonymises judicial decisions prior to publication. This system is fully integrated into the digital court platform through which judicial proceedings are currently processed, and it operates using AI.

Legal research, analysis and drafting support

The Mexican Supreme Court has developed JulIA, a tool designed to navigate the Court’s judicial decisions and support legal research. Unlike traditional case‑law search engines, which rely on identifiers, broad topics, or keywords, JulIA allows users to submit complex queries in natural language. The development of JulIA began in 2022, and the platform is still in a pilot phase as at July 2026. Although it is intended to become publicly accessible, as at July 2026 access remains restricted to Court personnel.

Defence

Legal research, analysis and drafting support

Criminal defence lawyers in Mexico have a range of AI tools available, which may assist them with legal research, analysis and drafting support. This includes tools such as Harvey, vLex, LexisNexis, and CoCounsel, and individual practitioner tools such as Litigia, Lexius, Juriscribe, and Sof-IA. Those platforms offer bundled features, including legal research, information organisation, analysis, and document generation.

Victims

Victims have direct standing in Mexican criminal proceedings. Under Article 20(C) of the Mexican Constitution and Articles 105 and 109(XV) of the National Code of Criminal Procedure, a victim or offended person can be formally recognised as a party and may participate throughout the proceedings, either personally or through legal counsel. This includes requesting investigative acts, challenging prosecutorial omissions, seeking reparation and appealing certain judicial decisions, even where the victim has not joined the prosecution as a co-complainant. Victims may also assist a public prosecutor pursuant to Articles 338 and 339 of the National Code of Criminal Procedure. Although criminal prosecution ordinarily remains the responsibility of the Public Prosecutor’s Office, Articles 426 and 428 of the National Code of Criminal Procedure exceptionally permit victims to bring a private criminal action directly before the supervisory judge for certain complaint-based offences carrying limited penalties.

The development of AI tools for use by victims in Mexico represents a significant area of AI deployment.

Case management

The Offices of the Attorney General of the States of Zacatecas and Coahuila, in partnership with LAB‑CO, created Ebba, a citizen‑facing chatbot. This tool is designed to improve the user experience for citizens who need to interact with Mexican prosecutors, for example, when filing a criminal complaint. It provides general information about the prosecutors’ offices and guides users through judicial and administrative procedures. Ebba is available on WhatsApp, Facebook Messenger, Telegram, and the official website. During the first six months after deployment, the chatbot of the Attorney General’s Office in Coahuila alone was used by more than 4,000 citizens.

The Guanajuato State Prosecutor's Office launched Fiscal IA, an AI-powered system designed to automate the creation of citizens' reports (actas circunstanciadas). This is a routine but time-intensive legal procedure, in which citizens that have experienced crime relate the relevant facts to the public authorities. The tool operates through ‘smart cabins’ available around the clock, allowing users to complete the process without requiring public servants' assistance at every step. The tool has reduced the average time taken to complete a citizen report from 35 minutes to 11. Given that circumstantial reports require the submission of personal information, the system was built to store all data on local servers rather than in the cloud, ensuring greater control over sensitive information.

Legal research, analysis, and drafting support

Two chatbots have been developed to help individuals understand judgments of the Mexican Supreme Court:

1. JusticIA (also known as Mati) is a publicly available chatbot launched by the Mexican Supreme Court in 2025. It functions as a separate website or as a pop-up chat in the Court’s main site and assists users with both basic general questions about the Court’s structure and procedures, as well as specific inquiries related to individual cases. It can identify rulings, summarise them, clarify aspects of the decisions or the litigation process, and direct users to the original text for easy reference. It accepts questions in natural language.
2. Sor Juana is a publicly available chatbot in pilot use, designed to help users better understand the judicial decisions of Mexico’s Supreme Court. This system is powered by an LLM and communicates in natural language. It can handle complex questions about judicial rulings and engage in a direct ‘dialogue’ about them. Users can ask about the case itself, the key arguments, the references cited, potential legal and policy implications, and more. At present, Sor Juana is in a pilot phase and has access only to the decisions initially proposed by Justice Ana Margarita Ríos Farjat during her tenure on the Supreme Court (December 2019 – August 2025), and later adopted by the court. As at July 2026, it remains uncertain whether the tool will continue to be updated with additional rulings.

Mexico’s Federal Electoral Tribunal has also adopted Alfonsin, a publicly accessible chatbot which facilitates access to and understanding of the Federal Electoral Tribunal’s rulings. The tool draws on information about electoral processes, citizen‑electoral rights, rulings, case law, appeals, and other election‑related topics, and communicates in natural language.

Moreover, in 2025, the Supreme Court of Justice of the State of Jalisco launched an AI tool on its website called Resumen Ciudadano de Sentencias, which operates as a generative-AI ‘judicial translator’. The tool analyses judicial decisions, identifies key concepts and complex structures, and produces a legally faithful plain-language summary—for example, turning 100–150-page rulings into an accessible version in seconds, helping non-lawyers to understand the judgment.

Separately, the Mexican NGO TOJIL, A.C. has also developed an anti-corruption chatbot called TEO, which can provide legal guidance to citizens who have experienced or witnessed corruption‑related crimes committed by public officials, including prosecutors and police officers. The advice offered through the chatbot is supported by the organisation’s lawyers, who contact users directly after their initial interactions with the AI‑powered system to provide more tailored and reliable guidance on where to file a complaint, how to follow up on it, and related matters. TEO operates in Mexico City, the State of Mexico, Nuevo León, Oaxaca, Puebla, and Quintana Roo. Since 2022, it has facilitated more than 214,000 conversations. As at July 2026, its continued operation across all six jurisdictions cannot be confirmed.

TRAINING

As at July 2026, there are no systematic or mandatory training programmes on the responsible use of AI available to law enforcement, prosecutors, judges, or lawyers in Mexico. Instead, the landscape is best described as a broad but decentralised set of initiatives led by different institutions with varying scope and depth.

Within prosecutorial and law enforcement institutions, training programmes generally focus on investigative methods, forensic science, intelligence analysis, and the operation of technological systems used in criminal investigations. For example, the Criminal Investigation Agency (Agencia de Investigación Criminal) within the Attorney General’s Office provides professional training related to investigative techniques and the use of technological tools supporting intelligence analysis. The Office of the Attorney General, through the National Institute of Criminal Sciences, has promoted institutional workshops on AI and related topics, including digital evidence in criminal prosecution.

With respect to judges:

1. Within the federal judiciary, professional training and continuing education are provided through the Federal School of Judicial Formation (Escuela Federal de Formación Judicial), which operates under the Federal Judiciary Council (Consejo de la Judicatura Federal) (CJF). Training programmes offered through this institution include instruction on digital judicial systems, electronic case management, and the use of the judiciary’s digital platforms, including the ‘expediente electrónico’ and other tools associated with the judiciary’s ‘Justicia Digital’ initiatives.
2. The Supreme Court of Justice of Jalisco held a workshop in November 2025 comparing different large language models and teaching participants how to draft effective prompts.
3. Mexico City’s judiciary organised seminars in 2025 on AI and disciplinary law.

For lawyers, the Mexican Bar Association has promoted upskilling in emerging technologies through a postgraduate diploma on generative AI, developed in collaboration with ioJustice, an independent legal-technology consultancy specialising in digital justice, technological development and professional education.

REGULATION

Mexico's regulatory approach to AI in criminal justice remains fragmented, relying on general legal frameworks rather than a dedicated AI statute. There is no comprehensive legislation governing AI in criminal proceedings; instead, the Federal Judiciary's e-Justice Policy sets principles of transparency and oversight for judicial digital tools, while general laws—the Federal Criminal Code, the National Code of Criminal Procedure, and Mexico's data protection statutes—may apply to AI-generated evidence, deepfakes, and personal data processing without addressing AI directly. There are no AI-specific guidelines for practitioners, though existing judicial and bar ethics codes may extend to AI-assisted work, alongside constitutional protections for due process, privacy, and access to justice, and Mexico's international human rights obligations.

AI Regulations

E-Justice Policy (Política de e-Justicia) (2022)

The Federal Judiciary Council (Consejo de la Judicatura Federal), the administrative and oversight body of Mexico’s Federal Judiciary, has adopted a ‘Política de e-Justicia’ which establishes principles for the development and governance of digital justice systems within the federal judiciary. The policy emphasises transparency in the use of technological tools, reliability and security of digital systems, and institutional oversight of algorithmic processes, including consideration of potential bias and discriminatory outcomes. The policy does not establish explicit prohibitions on the use of AI systems. However, it requires that digital tools used within the judiciary operate under principles of transparency, reliability, and institutional oversight, and that safeguards be considered to mitigate potential bias or discriminatory effects. The policy applies to judicial bodies and administrative units of the Federal Judiciary overseen by the CJF.

Deepfakes and synthetic evidence

As at July 2026, Mexico has not enacted legislation specifically regulating ‘deepfakes’ as a distinct legal category. There is no general statutory obligation requiring the labelling or disclosure of AI-generated audio, video, or image content. However, amendments to the Federal Copyright Law enacted in May 2026 specifically protect performers against certain AI-generated cloning or impersonation of their images, voices and performances.

The use and dissemination of manipulated digital content may, however, fall within existing criminal, civil, and constitutional frameworks, depending on the circumstances. Relevant provisions include offences under the Federal Criminal Code (e.g., fraud and falsification), data protection obligations under the Federal Law on Protection of Personal Data Held by Private Parties, and constitutional protections concerning freedom of expression and dissemination (Articles 6 and 7), private life (Article 6) and personal data (Article 16).

Guidelines for Practitioners

As at July 2026, no guidelines have been issued for practitioners specifically concerning the responsible use of AI. Instead, pre-existing general guidelines continue to apply to the use of AI in legal proceedings.

As at July 2026, Mexico has not formally adopted UNESCO’s Guidelines for the Use of AI Systems in Courts and Tribunals (2025).

Ethics Code of the Federal Judiciary (Código de Ética del Poder Judicial de la Federación) (2004)

The Ethics Code of the Federal Judiciary governs the conduct of members of the Federal Judiciary. The Code establishes principles including independence, impartiality, objectivity and professionalism in judicial conduct. Although the Code does not explicitly address AI, these principles may apply to the use of technological tools in judicial work by requiring judges to examine files thoroughly (section 4.4), provide reasoned decisions (section 4.5), preserve professional secrecy (section 4.9), personally perform non-delegable functions (section 4.10) and prudently obtain and assess information (section 5.3).

Bar Association Ethical Rules

The Mexican Bar Association’s (Barra Mexicana, Colegio de Abogados) Professional Ethics Rules, adopted in 2017, require technical competence and continuing education (sections 2.1-2.2), prohibits false assertions and knowingly false evidence (sections 3.2 and 9.5), requires diligence and supervision (sections 10.7 and 13.1), preserves the lawyer’s responsibility for legal work (section 21.4) and protects professional secrecy (sections 22-26). These obligations may require lawyers using AI-assisted tools for legal research or drafting to verify the accuracy of such outputs and ensure that confidential client information is not improperly disclosed.

Comparable ethics rules include the ANADE Code of Ethics⁠, which addresses diligence, probity, good faith and professional secrecy (Articles 2-3 and 10-12), and the INCAM Code of Ethics⁠, which addresses independence, integrity, professional secrecy, clients’ interests and continuing professional training (sections 2.1-2.3 and 2.7-2.8). Neither appears to contain provisions specifically addressing AI.

Criminal procedure rules

National Code of Criminal Procedure (Código Nacional de Procedimientos Penales) (2016)

The National Code of Criminal Procedure (Código Nacional de Procedimientos Penales), which governs criminal proceedings in Mexico, does not expressly regulate the use of AI. However, its provisions governing investigative methods, admissibility of evidence, and procedural guarantees may apply to evidence generated or analysed using AI technologies. In particular, rules concerning legality of investigative methods, evidentiary integrity, and the right of parties to challenge evidence may affect the use of AI-assisted investigative tools or analyses in criminal proceedings.

Relevant provisions of the National Code of Criminal Procedure⁠ include: Articles 251 to 252, which distinguish investigative acts that may be conducted without judicial authorisation from those affecting constitutional rights and therefore requiring prior authorisation; Articles 263 to 265, which require evidence to be lawfully obtained and permit the exclusion or annulment of evidence obtained in violation of fundamental rights; and Article 337, which requires the parties to disclose the evidence they intend to present, including expert reports. Article 346 permits the exclusion of irrelevant, unnecessary, unlawfully obtained or otherwise inadmissible evidence, while Articles 356 to 359 require evidence to be properly introduced and evaluated freely and logically. These provisions apply equally where AI is used to obtain, generate or analyse evidence.

Deepfakes and synthetic evidence

In criminal proceedings, digital evidence must comply with evidentiary standards under the National Code of Criminal Procedure, including chain of custody, disclosure, and expert evidence requirements. There are no statutory standards specifically addressing deepfake detection or verification. The admissibility of allegedly manipulated media is therefore assessed under general principles of authenticity, reliability, and evidentiary weight. The National Code of Criminal Procedure also imposes disclosure obligations and permits challenges to the authenticity and reliability of evidence. The use of manipulated synthetic media may therefore raise due process concerns.

Articles 227 and 228 of the National Code of Criminal Procedure establish chain-of-custody requirements, including records of the evidence’s identity, original condition, collection, preservation, transfer and every person who handled it. A chain-of-custody irregularity does not automatically eliminate probative value, but materially altered evidence must be assessed together with other evidence. Articles 368 and 369 permit suitably qualified experts to examine matters requiring specialised knowledge and Article 372 allows the opposing party to cross-examine them. Article 383 permits material or documentary evidence to be introduced only after it has been authenticated, while Article 390 expressly allows rebuttal evidence when a dispute arises concerning the ‘veracity, authenticity or integrity’ of evidence. Accordingly, alleged deepfake evidence could be challenged through chain-of-custody records, expert examination, cross-examination and rebuttal evidence, although the Code contains no deepfake-specific verification standard.

The Federal Criminal Code criminalises the alteration, destruction, or concealment of evidence. The introduction of fabricated or manipulated digital material in judicial proceedings may constitute offences such as falsification, fraud, or obstruction of justice. Article 225(XXXI) criminalises a public official’s alteration, concealment or destruction of evidence or interference with the chain of custody, while Article 225(XXXII) criminalises a public official’s obstruction of a criminal investigation. Article 231 applies to lawyers and litigants and criminalises knowingly relying upon false facts or documents and altering evidence and presenting it in proceedings to obtain an unlawful decision. Articles 243 to 246 address the falsification and knowing use of public or private documents, and Article 386 applies where deception is used to obtain property or an unlawful financial benefit. Accordingly, presenting AI-generated or manipulated material is not automatically an offence, but may fall within these provisions where their specific requirements are established.

Data protection legislation

General Law on Protection of Personal Data Held by Private Parties (Ley Federal de Protección de Datos Personales en Posesión de los Particulares), 2025

The General Law on Protection of Personal Data Held by Private Parties (Ley Federal de Protección de Datos Personales en Posesión de los Particulares), regulates the processing of personal data by private entities in Mexico. Its provisions regarding lawful processing, data security, and the protection of sensitive personal data may apply to private technology providers developing or operating AI systems that process personal data relevant to criminal investigations or judicial proceedings. The law applies to private individuals and entities processing personal data through manual or automated procedures (Articles 1 and 2). Articles 5 and 6 require lawful, fair, purpose-specific and accountable processing. Articles 7 to 9 regulate consent, including express written consent for sensitive personal data. Articles 10 to 12 require accuracy, deletion when data are no longer necessary, purpose limitation and data minimisation and Articles 14 to 16 require an appropriate privacy notice. Article 18 requires administrative, technical and physical safeguards against loss, alteration, destruction and unauthorised use, access or processing. Article 19 requires prompt notification of significant security breaches and Article 20 imposes continuing confidentiality obligations. Articles 21 to 26 establish rights of access, rectification, cancellation and objection, including a right to object to certain exclusively automated processing, while Articles 35 and 36 regulate domestic and international transfers. These requirements may apply to an AI provider either as the entity responsible for processing or as a processor acting on another entity’s behalf.

General Law on Protection of Personal Data Held by Obligated Subjects (Ley General de Protección de Datos Personales en Posesión de Sujetos Obligados) (2025)

The General Law on Protection of Personal Data Held by Obligated Subjects (Ley General de Protección de Datos Personales en Posesión de Sujetos Obligados) governs the processing of personal data by public authorities, including justice institutions. It establishes obligations relating to the protection, security, and lawful use of personal data held by government entities, which may affect the use of AI systems processing personal information within criminal justice institutions. Articles 23 and 24 require accountability measures, including internal policies, staff training, audits and data protection by design and by default. Articles 25 to 29 require administrative, physical and technical security measures, risk and vulnerability assessments, a security document and notification of significant breaches. Articles 52 to 58 regulate external processors and cloud services, including contractual controls, confidentiality, deletion and restrictions on unauthorised access. Specifically for security, prosecution and justice institutions, Articles 74 and 75 limit collection and processing to data that are necessary and proportionate to their statutory functions, while Article 76 requires ‘high-level’ security measures protecting integrity, availability and confidentiality. These requirements apply where such institutions use AI systems to process personal data.

Cybersecurity laws

National Cybersecurity Strategy (Estrategia Nacional de Ciberseguridad) (2017)

Mexico’s National Cybersecurity Strategy (Estrategia Nacional de Ciberseguridad) establishes national policy objectives for protecting digital infrastructure and strengthening cybersecurity governance. Although not specific to AI, the strategy reflects broader policy considerations regarding the secure management of digital technologies used by government institutions, including those in the justice sector. Adopted in 2017, the strategy is a policy instrument rather than binding legislation. Its five strategic objectives concern society and rights, the economy and innovation, public institutions, public security and national security. Objective three calls for protection of government information and computer systems and continuity of public services. Relevant pillars within the strategy include ‘Standards and Technical Criteria’, which promotes common standards and security methodologies; ‘Critical Infrastructure’, which seeks risk reduction and continuity of essential services; and ‘Legal Framework and Self-Regulation’, which calls for harmonisation of cybercrime legislation and greater legal certainty. These objectives are relevant to justice institutions’ digital and AI systems, but the strategy itself does not impose legally enforceable security requirements.

As at July 2026, Mexico had no single comprehensive federal cybersecurity law in force. However, Chapter II of the Federal Criminal Code, entitled ‘Unlawful Access to Computer Systems and Equipment’, criminalises specified conduct. Article 211 Bis 1 covers unauthorised copying, modification, destruction or loss of information in computer systems protected by a security mechanism. Articles 211 Bis 2–3 establish more serious offences involving government and public-security systems, including conduct by persons who already have authorised access; penalties may be doubled where the conduct obstructs the administration of justice or affects criminal-proceeding records. Articles 211 Bis 4–5 contain corresponding offences for systems belonging to financial institutions, and Article 211 Bis 7 increases penalties where the information obtained is used for the offender’s or another person’s benefit. These provisions criminalise unauthorised interference but do not impose a general obligation upon every organisation to implement cybersecurity controls. Such affirmative security duties arise principally under data-protection legislation, including Articles 18 to 20 of the General Law on Protection of Personal Data Held by Private Parties and Articles 25 to 29 and 76 of the General Law on Protection of Personal Data Held by Obligated Subjects.

Human rights

The Mexican Constitution guarantees fundamental rights relevant to the use of technological tools in criminal proceedings, including the rights to due process, privacy, equality before the law, and access to justice. These constitutional protections may operate as limits on the use of AI technologies where such tools affect fairness, transparency, or the protection of personal data in criminal proceedings. In particular, Article 1 prohibits discrimination and requires all authorities to promote, respect, protect and guarantee human rights, Article 6 protects access to information and personal data, Articles 14 and 16 establish legality, due-process and privacy protections, Article 17 guarantees access to prompt, complete and impartial justice and Article 20 sets out the procedural rights of accused persons and victims in criminal proceedings. Where children are involved, Article 4 requires the best interests of the child to guide all State decisions and actions.

Mexico is also party to international human rights treaties applicable to the use of technological tools in criminal justice systems. The American Convention on Human Rights protects judicial guarantees under Article 8, privacy under Article 11, freedom to seek and receive information under Article 13, the rights of the child under Article 19, equality before the law under Article 24, and effective judicial protection under Article 25. The International Covenant on Civil and Political Rights contains corresponding protections concerning a fair hearing and criminal due process under Article 14, privacy under Article 17, access to information under Article 19, and equality and non-discrimination under Article 26. Where children are involved, Articles 2, 3, 12, 16 and 40 of the UN Convention on the Rights of the Child respectively protect non-discrimination, the child’s best interests, the right to be heard, privacy, and specific guarantees in juvenile criminal proceedings.

Outlook

As at July 2026, Mexico does not have a comprehensive national strategy addressing the use of AI in criminal proceedings. Existing digital and innovation policies operate at a general level and do not provide a coordinated framework for the deployment or oversight of AI within the justice system. There is, however, a growing recognition of the need to adapt existing legal frameworks to emerging technologies. Key concerns include due process, transparency, and the evidentiary treatment of AI-assisted outputs, particularly given that current procedural rules were not designed with such technologies in mind.

Legislative and policy proposals concerning AI have been developed in recent years. In May 2024, the Senate presented the non-binding National Artificial Intelligence Agenda for Mexico 2024–2030, which proposed general policy priorities for the development and governance of AI. In July 2025, the Senate’s Commission for the Analysis, Monitoring and Evaluation of the Application and Development of Artificial Intelligence published a proposed regulatory framework contemplating a General Law to Regulate and Promote Artificial Intelligence in Mexico. The Commission described its 2024–2025 work as a technical legislative process intended to establish a route towards the construction of a comprehensive regulatory framework.

A separate bill entitled the National Law to Regulate the Use of Artificial Intelligence was introduced in the Senate in February 2026. It would establish a general framework applicable to the public and private sectors, create a specialised regulatory agency and impose obligations based partly on the risks presented by AI systems.

As at July 2026, however, these measures remained proposals rather than binding legislation. Official Senate statements describe the regulatory framework as being under construction, and no proposal has yet established a framework specifically governing the use of AI in criminal proceedings. Accordingly, no definite timetable for adoption of comprehensive AI legislation has been announced.

CASES

Though there have been no reported cases concerning the use of AI in Mexican criminal proceedings as at July 2026, the issue has arisen in a proceeding originating from a civil property dispute. In Queja 212/2025, the executors of an estate, claiming to be third parties unrelated to the underlying proceedings, sought indirect amparo relief to protect two parcels of land that they asserted formed part of the estate. They requested that the amparo claim be recorded in the relevant real property registries, and the District Court required security as a condition of that registration. The executors challenged the amount of the security on the ground that it had been determined without objective parameters. In resolving that challenge, the Second Collegiate Court in Civil Matters of the Second Circuit used AI tools solely to assist with the numerical calculation of the appropriate security; AI was not used to determine ownership of the properties or otherwise decide the substantive merits of the dispute. The proceeding gave rise to two related judicial criteria: Tesis II.2o.C.9 K (11a.), Registro digital 2031009, recognising AI as a permissible auxiliary tool for calculating security in amparo proceedings, and Tesis II.2o.C.8 K (11a.), Registro digital 2031010, establishing minimum safeguards for its ethical and responsible use, including proportionality, protection of personal data, transparency and explainability, and human supervision and decision-making.