Alice — Brazil's AI system for detecting procurement fraud

Alice (Análise de LICitações e Editais) was created by Brazil's Comptroller General of the Union (CGU) in 2015 to continuously monitor federal procurement notices across Compras.gov.br, Licitações-e and the Diário Oficial da União, using natural-language processing and robotic process automation to detect irregularities across roughly 40 predefined risk typologies.

The system processes more than 500 procurement notices per day, sending alerts to over 500 users across CGU, the Federal Court of Auditors (TCU, since 2016), the Judiciary, the Public Prosecutor's Office, and state and municipal comptrollers.

Between 2019 and 2022, Alice-triggered alerts led to the suspension or cancellation of bids totalling R$9.7 billion, including R$504 million in losses avoided in 2021 alone; preventive audits enabled by Alice are completed in 8 days versus more than 400 days for traditional evaluations. In 2023 the system analysed 190,923 acquisitions, supporting audit work valued at R$27 billion. A peer-reviewed 2024 study (Springer, AI & Society) found Alice alerts associated with a nearly 20% increase in total procurement, consistent with reduced risk-aversion in contracting decisions under AI oversight.

OECD OPSI documents Alice as a global public-sector innovation case; known limitations include reliance on structured data from official portals, coverage limited to federal spending visible in those systems, and false positives that require human review.