On June 24, 2025, the Brazilian federal government published Ordinance No. 961/2025 (Ordinance) establishing, for the first time in the country, guidelines for the use of AI in criminal investigations and public security intelligence operations. It became effective immediately upon its publication, aiming to update the practices of Brazil’s law enforcement agencies while keeping a firm commitment to protect citizens’ fundamental rights.

This Ordinance was issued by the Brazilian Minister of Justice and Public Security (MJSP, Ministério da Justiça e Segurança Pública), an agency within the federal government’s executive branch. MJSP works to ensure the integrity of the Brazilian legal system and to protect the country’s political and constitutional rights. The agency also plays a key role in combating drug trafficking, corruption, money laundering, and the financing of terrorism.

What’s in the Ordinance

MJSP based the Ordinance on three fundamental pillars:

1. Strengthening investigative and intelligence mechanisms through advanced technological solutions.
2. Robust safeguards of personal data to prevent unlawful use in fraud, financial scams, and harassment.
3. The establishment of clear parameters for the application of AI in public security operations.

Among its provisions, the Ordinance authorizes the use of the technology in criminal investigations provided that it is used according to the law and following the principles of justice of due process, reasonableness, and a clearly defined purpose. MJSP affirms that the regulation aims to establish an ethical standard to ensure efficiency without compromising citizens’ fundamental rights. It also makes clear that the improper use of technology solutions, as defined by the Ordinance, may result in administrative, civil, and criminal liability.

Nonetheless, the Ordinance allows the use of AI that enables real-time remote biometric identification in publicly accessible spaces in the following cases:

1. Criminal investigations where there is reasonable evidence of authorship or involvement in a criminal offense.
2. The search for victims, missing persons, or situations involving serious threats to life or physical integrity of individuals.
3. Situations of individuals caught in the act of committing a crime.
4. The capture of fugitives.
5. The execution of arrest warrants.

In the domain of correction facilities, authorities will be permitted to use AI technologies to detect, locate, and block signals from mobile devices, as well as to access information stored on such devices. Regarding the latter, the Ordinance stated that it will be allowed only when they are lawfully seized.

All Brazilian federal security agencies, including the country’s Federal Police and the National Security Agency, are subject to the guidelines of the Ordinance. Likewise, state, district and municipal agencies that receive funding from the country’s National Public Security Fund. It also includes the Brazilian antitrust (CADE, Conselho Nacional de Defesa da Concorrência) and data protection (ANPD, Autoridade Nacional de Proteção de Dados) authorities, as applicable.

Ordinance follows a global trend of governments implementing guidelines and regulations on the use of AI by public authorities in the context of their responsibilities. It is no surprise that there is a growing concern over personal data and the misuse of these emerging technologies, which present new challenges on a daily basis.

DOJ AI and Criminal Justice Final Report

In December 2024, the US Department of Justice published the AI and Criminal Justice Final Report (DOJ Report), identifying key areas where AI can enhance law enforcement while ensuring the protection of privacy, civil rights, and civil liberties. It also addresses persistent operational and ethical challenges associated with the use of AI that extend beyond the current regulatory framework. According to the DOJ Report, there are four primary applications of the technology in the criminal justice system, which are (i) identification and surveillance, (ii) forensic analysis, (iii) predictive policing, and (iv) risk assessment.

The benefits of using AI for biometric identification are highlighted in the DOJ Report. However, it also raises concerns about the proper use of collected personal data and the accuracy of that information. Also, it emphasizes the risks associated with using AI in data analytics to forecast crime patterns, particularly the potential to reinforce systemic biases.

Additionally, the DOJ Report offers a critical recommendation regarding human oversight. It states that AI should support, but not replace, human decision-making. The report also calls for agencies to establish clear guidelines for the use, oversight, and compliance of AI systems, as well as to disclose information about AI usage, safeguards, and governance.

Correspondingly, in April 2025, the White House Office of Management and Budget (OMB) released new policies on Federal Agency Use of AI and Procurement, OMB Memorandum M-25-21 and OMB Memorandum M-25-22. These guidelines aim to encourage and promote American AI innovation by offering clear direction on AI adoption, eliminating unnecessary bureaucratic barriers, and enabling US agencies to operate more efficiently and cost-effectively.

In line with the DOJ Report, the new policies also require agencies to provide human oversight and intervention where appropriate, as well as to offer remediation measures for individuals eventually affected by AI-enabled decisions.  

Effective compliance

It is evident that governments, as well as private entities, are increasingly leveraging AI to enhance efficiency and reduce operational costs. However, adopting such technologies without a full understanding of their scope or their implications within existing legal frameworks poses significant risks, particularly in today’s fast-paced, evolving technological and regulatory landscapes.

An effective compliance program is essential to navigate these new and complex scenarios. At least, it should include thorough risk assessments, ongoing human oversight, and responsive remedial measures to build a program grounded in both legal and technical expertise. This approach is crucial to employ the benefits of AI while safeguarding the integrity and sustainability of operations across both public and private sectors.