Chile

USE

Law enforcement 

Predictive analytics 

The Security Analysis and Mathematical Modeling Center at the University of Chile, working with Carabineros' Criminal Analysis Department, developed a system known as the ‘Urban Crime Prediction System’. First piloted in 2015, the tool uses historical crime data, geographic information, and temporal patterns to predict where crimes such as robberies and assaults are most likely to occur. The system combines three predictive algorithms to generate colour-coded risk maps that indicate hotspots by time and location, and these maps are integrated into Carabineros’ existing criminal-analysis platform for patrol planning and resource allocation. 

Unlike many similar programmes worldwide, the Chilean software not only seeks to predict crime but also enables search and filtering by specific crime type, for example, robbery with force, robbery with violence, or sexual assault and can link these crimes to particular activities and services, such as liquor stores, banks, and service stations, which often become focal points for criminal activity. 

Initially tested in 12 communities of Santiago, it later expanded to more than 70 municipalities, with early reports claiming up to 89% accuracy in identifying risk areas.

Data review and analysis 

The ‘Sistema de Teleprotección Nacional’ (SITIA) is a government-led tele-protection platform that integrates facial recognition, automatic license-plate reading, and AI-assisted analytics with existing public and private camera infrastructure. As at March 2026, SITIA operates in the Región Metropolitana and Región de Magallanes, and is being expanded to the Región de Biobío. Chile’s Ministry of the Interior and Public Security has also launched the first municipal pilot of SITIA in the municipality of Lo Barnechea, which has connected 115 license-plate cameras with the Carabineros national stolen vehicle database.

The SITIA system interconnects more than 2,000 security cameras in the Metropolitan Region of Santiago, Chile, in addition to over 300 cameras from the Traffic Control Operations Unit. It is designed to prevent and respond to vehicle theft, locate fugitives and missing persons, and generate real-time alerts to Carabineros through AI-enabled license plate recognition. In Magallanes, a proof-of-concept facial recognition system has reportedly generated over 64,000 detections. According to government authorities, SITIA supports coordinated data-sharing between municipalities, Carabineros, and the Public Prosecutor’s Office.

The system is designed to detect missing persons or individuals with outstanding arrest warrants through facial recognition and to identify stolen vehicles via automatic license-plate reading. According to the government, SITIA will support both the Carabineros and the Public Prosecutor’s Office in the investigation and prosecution of crimes. 

The Calle Segura program, launched in 2019, was an initiative to deploy over 2,000 cameras with advanced technology and AI across 1,000 critical points nationwide. Part of this included drones for televigilance, checkpoints to intercept escape routes for stolen vehicles, and improved interconnection between law-enforcement agencies and camera systems. The system connects various institutions—Carabineros, municipalities, highways, prosecutors—via a platform to share surveillance data and analytical outputs. The ‘Calle Segura’ programme also includes the implementation of a ‘comisaría virtual’ (virtual police station) to facilitate procedures that previously required in-person visits, thereby freeing Chilean national police to increase their presence on the streets. 

Prosecutors  

Evidence review and analysis 

The Chilean Attorney General’s Office uses an AI system called ‘FISCAL HEREDIA’, now used by more than 240 prosecutors nationwide. Using large volumes of data obtained from the databases of the Prosecutor’s Support System and the ‘SOSAFE’ citizen-security application (a community reporting platform), the tool employs two primary models: 

1. Mapping connections between individuals with criminal records, aiding in identifying suspects and understanding criminal networks.
2. Using natural language tools to process victim statements and police reports to detect patterns in criminal behaviour, assisting in linking related cases.

All results generated by FISCAL HEREDIA are subject to human oversight and validation by prosecutorial teams before integration into cases.

To develop the project, both the Prosecutor’s Support System and the SOSAFE citizen-security application made their databases available to researchers, while complying with privacy protection standards.

In pilot testing, the system was able to identify multiple suspects linked to a known individual, some of whom were still at large. The tool can work continuously, scanning hundreds of thousands of police reports. While it does not replace the work of legal experts, it offers a powerful analytic tool to enhance and accelerate their prosecutorial investigations.

The development and deployment of the FISCAL HEREDIA tool had multiple objectives:

1. Enhance prosecutorial intelligence. The system links incidents, such as thefts committed using vehicles with the same license plate or similar characteristics reported through the SOSAFE application to help identify ‘crime drives’, or patterns of repeated criminal activity,
2. Predictive burglary models. It produces Predictive home burglaries risk forecasts in different areas for better management of neighbourhood security, to enable a more effective preventative patrol.  
3. Improve alert management. It improves the organisation of alerts sent through the SOSAFE application and received by police and municipalities, providing a better prioritisation of urgent events requiring a quick response. 

Looking ahead, the Prosecutor’s Office plans to expand the tool’s capabilities by integrating additional datasets, such as prison records and inmate associations, to further enrich the social network model. They also aim to refine how crimes are weighed within the system (for instance, factoring in the severity of sentences) and to host the model on a central server or cloud platform so it can be used nationwide by regional prosecutors and police. Eventually, it is hoped that the technology will be adapted to support investigations into other serious crimes such as homicide, fraud, kidnapping, and drug trafficking.

Courts 

In September 2025, Pablo Cabezas, head of the Innovation Laboratory of the Judiciary, said that Chile’s judiciary was in an “exploratory” phase of AI adoption, where it is learning how AI can serve the objectives of the judiciary and give judges better tools to address their workload.

Case management 

As at March 2026, a pilot tool named ‘Justa’ is being tested by a team of judges, administrators, and technical advisors at the Juzgado de Letras y Garantía de Mulchén (Court of Letters and Guarantees of Mulchén), which handles criminal cases among others. It aims to speed up judicial workflows by assisting with internal tasks such as transcribing audio, drafting summaries, generating hearing minutes, and processing legal documents. This system is expanding to other courts, including a chamber of the Supreme Court. 

Other pilots include automated transcription and document analysis in local courts, such as Mulchén.

Legal research, analysis and drafting support 

Chile’s Supreme Court uses ‘Buscador de Jurisprudencia’ (Jurisprudence Search Tool), established by Auto Acordado No. 164-2024 (effective January 2025) and designed as a nationwide search engine for court ruling. The tool uses AI for semantic search anonymisation and relevance ranking, aiding judges and clerks in legal research.

TRAINING

Chile has begun incorporating training programs and awareness initiatives for justice-sector professionals on the use and risks of AI. The Academia Judicial de Chile (a public institution providing training to legal practitioners) has run ethics-focused sessions on AI and announced that it is developing guidance and training instruments to promote a ‘responsible, realistic’ use of AI in courts, with an emphasis on transparency, bias, and data protection. In 2022, the Attorney General’s Academy promoted and hosted AI-focused activities to familiarise prosecutors with new tools such as FISCAL HEREDIA. 

At the academic level, the Universidad de Chile launched a short course in 2025 titled ‘Inteligencia Artificial Aplicada a la Justicia’ (Artificial Intelligence Applied to Justice), open to practitioners and officials, covering applications of AI in legal processes.  

In August 2025, the Colegio de Abogados de Chile (Chilean Bar Association) convened a national seminar on ethics and AI, bringing together members of the judiciary and practising lawyers to debate risks, responsibilities, and ethical limits. 

Training on AI use  is also available in Chile through UNESCO-specific  programmes. 

As at March 2026, there is no publicly available information indicating that Chilean judges receive specific training on the detection of deepfakes or other forms of synthetic media

Regulation

As at March 2026, there is no regulation expressly governing the use of AI in criminal proceedings or in judicial proceedings more broadly. However, existing regulations—such as the Criminal Code of Procedure, data protection laws and constitutional principles—may be construed to regulate or limit the use of AI in criminal proceedings.

In addition, a comprehensive AI draft bill (Boletín Nº 16.821-19) was approved by the Chamber of Deputies in October 2025 and has since been dispatched to the Senate, where it is under review. The draft AI bill proposes a risk-based framework for AI systems. The bill would introduce transparency, documentation, and oversight obligations for high-risk AI systems if enacted. The bill also reflects Chile’s broader engagement with international AI governance frameworks. In particular, it has been developed in alignment with UNESCO’s Recommendation on the Ethics of Artificial Intelligence, and builds on Chile’s prior application of UNESCO’s Readiness Assessment Methodology (RAM). While the bill represents a significant step towards a comprehensive AI regulatory framework, its final adoption will depend on the outcome of Senate deliberations and any subsequent amendments. The legislative timeline therefore remains uncertain.

In the absence of AI-specific legislation, the use of AI in criminal proceedings remains subject to existing legal and institutional safeguards. Constitutional guarantees—including due process and the right to privacy under Article 19 of the Chilean Constitution—may be invoked before the ordinary courts and the Constitutional Court. In addition, general oversight mechanisms, including judicial and prosecutorial disciplinary regimes, continue to apply. Data protection obligations relevant to AI systems will also fall within the supervisory and sanctioning powers of the Personal Data Protection Agency established under Law No. 21.719, once in force.

Guidelines for practitioners 

Judicial guidelines 

While there are no judicial guidelines specifically addressing the use of AI within the judiciary, judges are bound by constitutional principles of impartiality, independence, and reasoned decision-making. These imply caution in relying on automated tools and require that any use of technology be subject to appropriate human oversight.There is no publicly available information indicating that Chile has formally adopted UNESCO’s Guidelines for the Use of AI Systems in Courts and Tribunals.

However, certain principles reflected in those Guidelines—such as transparency, accountability, human oversight, and the mitigation of bias—are increasingly present in broader judicial training initiatives and policy discussions in Chile, particularly those led by the Academia Judicial. These initiatives focus on the responsible and ethical use of AI in the justice system, although they do not amount to a formal or systematic implementation of the UNESCO Guidelines within the judiciary.

There are also no specific guidelines for lawyers regarding the use of AI, as at March 2026.

Criminal procedure rules

Criminal Code of Procedure  

Pursuant to Article 297 of the Criminal Code of Procedure, judges must evaluate all evidence—which would include outputs from forensic software, facial recognition systems, or predictive analytics tools—in accordance with principles of logic, maxims of experience, and scientifically well-founded knowledge. Courts are also required to explain the reasoning behind their assessment. This evidentiary rule functions as a natural safeguard against admitting or relying on opaque or insufficiently validated AI systems that cannot meet scientific rigour or transparency thresholds.

Chilean law does not contain specific provisions expressly regulating the disclosure of AI systems as such in criminal proceedings. Disclosure obligations are governed by the Criminal Code of Procedure (Law No. 19.696), which regulates access to investigative records (Article 182), pre-trial disclosure of evidence and witness lists (Articles 259–260), and the admissibility and exclusion of evidence (Article 276). These provisions require the Public Prosecutor’s Office to disclose the evidence it intends to rely upon at trial and guarantee the defendant’s right to challenge its admissibility and reliability. Where outputs generated or assisted by AI systems are introduced as evidencefor example, through forensic software, predictive analytics, or facial recognition tools – they must comply with the general evidentiary standards set out in Article 297 of the Criminal Code of Procedure, which requires courts to evaluate evidence in accordance with logic, maxims of experience, and scientifically grounded knowledge and to provide reasoned justification for their conclusions.

As at March 2026, Chile has not enacted legislation specifically regulating deepfakes as a distinct legal category. There is no general statutory obligation requiring individuals to label or disclose synthetic or AI-generated audio, video, or image content as such. Although there is no “deepfakes” offence, harmful uses of AI-generated synthetic media may nevertheless fall to be addressed by other, existing criminal offences, including: (1) computer-related fraud and data interference under Law No. 21.459, where synthetic media is used to manipulate data or deceive systems; (2) false accusations (Article 211 of the Criminal Code), where synthetic content is used to knowingly make a malicious accusation before authorities; and falsification-related offences (Articles 193–197 of the Criminal Code), where fabricated digital material is introduced as authentic evidence in judicial proceedings; and (3) data protection violations under Law No. 21.719, where biometric or personal data is unlawfully processed.

Data protection legislation 

General Data Protection Law (Law 21.719) 

Law No. 21.719 on the Protection of Personal Data (LPDP), which was approved by the Chilean Congress in December 2024 and will come into full effect in December 2026, establishes a legal framework for personal data protection. The LPDP creates a Personal Data Protection Agency to ensure compliance, expands data subject rights like data portability and erasure, and imposes new obligations on data controllers, aligning Chile's data privacy rules with international standards. The LPDP applies to data controllers (entities or individuals who determine the purposes and means of processing personal data) and data processors (entities or individuals who process data on behalf of a controller) at both private and public entities (including courts and prosecutorial bodies). While not specific to AI, the LPDP regulates the processing of personal data, which is highly relevant to AI systems used in the justice sector (e.g., risk assessment tools, predictive policing). Obligations around lawfulness, transparency, proportionality, and data subject rights apply equally to AI systems in the justice sector. These could, for example, restrict the deployment of AI tools that involve biometric or other sensitive data, or require procedural safeguards to ensure fair treatment of defendants. 

Cybersecurity Laws

Law No 21.459 on Computer Crimes (2022)

Law No. 21.459 modernised Chile’s criminal framework on cybercrime and aligned domestic legislation with the Budapest Convention on Cybercrime. The law criminalises, inter alia, unlawful access to computer systems, illegal interception of data, interference with data or computer systems, misuse of devices, offences under Law No. 21.459, including computer-related fraud and unlawful interference with data or systems, and offences involving the misuse of credentials or authentication mechanisms.

Although Law No. 21.459 does not refer expressly to artificial intelligence systems, its provisions are technology-neutral and apply to conduct affecting the confidentiality, integrity, or availability of data and computer systems, including conduct carried out through automated or AI-assisted tools. In the context of criminal proceedings, these offences may be engaged where AI systems are used to unlawfully access, alter, intercept, or manipulate digital information or evidentiary material.

The law also provides procedural tools for the investigation of cybercrime, including provisions relating to digital evidence preservation and cooperation mechanisms, which may become relevant where AI-assisted systems are involved in cyber-enabled offences.

Human rights 

The use of AI in criminal proceedings must be consistent with procedural guarantees and fundamental rights included in the Chilean Constitution, including due process, equality before the law, and the right to a fair trial. Article 19 of the Constitution, which protects privacy, honour, and the inviolability of communications, as well as the right to data protection, may, for example, operate as broad constitutional constraints on the use of AI where sensitive personal or biometric data are processed, or where automated outputs may affect due process and equality before the law.

Relevant guarantees一including the right to a fair trial and privacy rights一under other international human rights treaties to which Chile is a party, such as articles 8 and 11 of the American Convention on Human Rights, articles 14 and 17 of the International Covenant on Civil and Political Rights or articles 16 and 40 of the Convention on the Rights of the Child, may also provide additional guidance.

Chile also participates in Inter-American Commission on Human Rights (‘IACHR’) initiatives concerning digital rights and the use of technology in justice systems. The IACHR has emphasized that any AI use in judicial contexts must safeguard access to justice and non-discrimination. Chile has also endorsed the UNESCO Recommendation on the Ethics of Artificial Intelligence (2021), which, while non-binding, sets principles on transparency, accountability, and human rights protections that extend to justice-sector applications.

Outlook 

In 2021, Chile launched its first National Artificial Intelligence Policy (2021–2030) which outlines the country’s vision for AI development in relation to three areas: (i) enabling infrastructure, (ii) AI development and adoption, and (iii) ethics, regulation, and socioeconomic impacts. The policy establishes the foundation for responsible use of AI across both the public and private sectors, including those in the justice sector

In May 2024, Chile presented an updated National AI Policy and accompanying draft law following UNESCO’s Readiness Assessment Methodology. The updated policy acknowledges the need for AI regulation, including guidelines on the ethical use of AI systems in the public sector, standards for transparency, accountability and data protection as well as the reform of laws complementary to AI. The draft law (Boletín No 16.821-19) proposes a regulatory approach that combines self-regulation with risk regulation, distinguishing AI systems into those presenting unacceptable risk, high risk, limited risk, and no-obvious risk. It mandates transparency, explainability, documentation, human oversight, and sanctions for non-compliance.

Cases

Biometric data

In a 2024 decision (Rol N° 18.566-2024), the Supreme Court held that Worldcoin一a digital ID platform co–founded by Open AI’s CEO Sam Altman一violated a minor’s constitutional rights to privacy and data-protection by scanning her iris without informed consent and ordered Worldcoin to delete her biometric data within 30 days. The Supreme Court highlighted that biometric data is highly sensitive, requiring special constitutional protection. The ruling leaves open important questions about how courts will handle biometric, AI-generated and AI-processed data in criminal proceedings.

Judicial use of AI

In 2020, the Chilean Constitutional Court addressed the constitutionality of conducting criminal trials via videoconference under pandemic-era legislation. The petitioner challenged the law that allowed remote trials, arguing it violated his right to a fair trial, including the right to an effective defence. The complaint was ultimately rejected (Rol N° 9054-2020) because the court did not reach the majority vote required to declare provisions of the law unconstitutional. Several justices, however, raised concerns about the use of digital and AI tools to replace judicial functions and the erosion of fair trial guarantees. In particular, the reasoning emphasised that the use of technological mechanisms in criminal proceedings must preserve due process and defence guarantees, and must not undermine “las garantías del debido proceso” or the effective exercise of the right of defence. Also, certain opinions also cautioned against the substitution of judicial functions, highlighting that AI should ‘never’ replace judicial decision-making (‘the judge’s subjectivity’) and evidence assessment, especially in criminal cases. These observations reflect broader constitutional sensitivities that may become relevant in future litigation involving AI-assisted tools.

Misuse of AI in court filings

In 2026, a Chilean civil court sanctioned a lawyer for submitting a judicial filing containing fabricated legal citations generated using AI (Cruz v. Manzano y Asociados, Rol C-6127-2025, 21 January 2026). The Segundo Juzgado Civil de Concepción imposed a fine after finding that the submission included non-existent case law and erroneous legal reasoning, in breach of the duty of procedural good faith. The court emphasised that the use of AI tools does not absolve legal practitioners of their responsibility to verify the accuracy of submissions. While arising in a civil context, the case is indicative of the risks associated with the use of AI in legal proceedings and underscores the continued application of existing professional and procedural standards to AI-assisted work.

Deepfakes and synthetic media

As at March 2026, there are no reported instances of deepfake-generated evidence being adduced in criminal proceedings.