evidoria

← Back to browse

Good practice Imported

Coaf Cloud — Brazil's Financial Intelligence Unit Plans to Add AI as Suspicious-Activity Filings Keep Climbing

Brazil · Brasília · See the Brazil profile · See the Brasília profile

Evidence: Descriptive / self-reported Top 83% 40/100 · Ask Evidence Copilot about this practice

Brazil's financial intelligence unit, Coaf, logged 18,762 financial-intelligence reports and 2.57 million suspicious-activity communications in 2024 (up 766.6% since 2015); from 2026 it plans to add AI to a new 'Coaf Cloud' system, with ethics guidelines still under discussion.

18,762 reports
Financial Intelligence Reports produced (2024)
2,566,713 communications
Suspicious-operation communications received (2024)
296,183 communications
Suspicious-operation communications received in 2015 (2015)
766.6 %
Increase in suspicious-operation communications since 2015 (2015–2024)
Coaf Cloud — Brazil's Financial Intelligence Unit Plans to Add AI as Suspicious-Activity Filings Keep Climbing

Details

Promoter
Conselho de Controle de Atividades Financeiras (Coaf), Brazilian Ministry of Finance
Period
2025–2026
Keywords
financial intelligence, anti-money laundering, public security cooperation

Context

Brazil's Conselho de Controle de Atividades Financeiras (Coaf), the country's financial intelligence unit, receives an average of about 7.5 million communications a year from banks and other obligated entities, of which only around 2% are classified as atypical. From this flow, Coaf produced 18,762 Financial Intelligence Reports (RIFs) in 2024 for use by police and prosecutors, while the number of suspicious-operation communications it received reached 2,566,713 that year — a 766.6% increase over the 296,183 recorded in 2015, according to Brazilian press analysis of Coaf's own data.

Activities

To keep up with this growth, Coaf — under the National Public Security Secretariat and in cooperation with the Central Bank — is migrating its systems to a cloud environment, "Coaf Cloud", for greater interoperability, and is being reinforced with analysts on loan from state governments. From 2026 it plans to add artificial intelligence to the pipeline, aiming to improve pattern and anomaly identification, automate cross-referencing of data from banks, fintechs, public bodies and private data providers, and speed up the production of qualified RIFs.

Conclusions

As of the sources reviewed (October 2025 and subsequent reporting), specific ethical guidelines for the algorithmic monitoring of transactions were still under discussion, intended to align Brazil's approach with Financial Action Task Force (FATF/GAFI) recommendations. The 2024 figures above describe Coaf's largely human-driven process before the AI system's planned 2026 launch; no independently evaluated results attributable to the AI component itself are yet available.

Implementation

Indicative cost
Medium (€50k–€500k)
Time to results
Medium (1–3 years) — Cloud migration ("Coaf Cloud") underway; AI component planned for addition from 2026, alongside ongoing analyst reinforcement from state governments.
Staffing & skills
Conselho de Controle de Atividades Financeiras (Coaf), under Brazil's National Public Security Secretariat, in cooperation with the Central Bank, Analysts on loan from state governments reinforcing Coaf's capacity

Conditions for success

  • Migration to a cloud environment ("Coaf Cloud") for greater interoperability across data sources (banks, fintechs, public bodies, private data providers)
  • Alignment of algorithmic-monitoring ethics guidelines with FATF/GAFI recommendations, still under discussion as of the sources reviewed

Common failure modes

  • Specific ethical guidelines for algorithmic transaction monitoring were still under discussion, not yet adopted, as of the sources reviewed
  • 2024 figures describe the pre-AI process; no independently evaluated results for the AI component itself are yet available

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful