At 7:26 on Friday, 6 December, Raphael Correa Druciak, 42, was filmed and scanned by a facial recognition camera while crossing a corner of Indianópolis Avenue, in São Paulo’s southern region. Druciak had been on the run for three months after being convicted of two homicides in 2002 in the metropolitan area. The city government also alleges he is a member of the Red Command (CV). To defend its controversial Smart Sampa program, the Ricardo Nunes (MDB) administration has highlighted arrests like this. The project, integrating AI-powered surveillance cameras, is the city’s flagship public security initiative.
This Content Is Only For Subscribers
To unlock this content, subscribe to INTERLIRA Reports.
Criticism of the Program
Criticism of the program emerged during its early stages. The first version of the Smart Sampa notice stated that skin color and cases of vagrancy could be used as criteria to identify suspects in the images. The city government attributed this to a translation error, and the issue was corrected before the bidding phase. The Municipal Court of Auditors (TCM) initially suspended the process due to concerns over potential violations of the General Data Protection Law and the rights of minorities, including the black population. However, the TCM eventually allowed the program to proceed after receiving clarifications from the city government.
Numbers So Far
By 4 January, the program had completed six months of operation. As of Monday (06/01), at least 1,400 people had been arrested for crimes identified through camera surveillance, and 28 missing persons had been located using facial recognition technology. During this period, 354 fugitives were captured, with a significant increase in arrests beginning in November following technical adjustments to the system. According to Junior Fagotti, Deputy Secretary of Urban Security and one of the program’s creators, there have been no recorded cases where an individual was wrongly identified, approached by authorities, and later found to be a different person from the system’s match.
Problems with Similar Systems
The Smart Sampa system only issues alerts when it is at least 90% confident in a match. This caution is necessary, as flaws in facial recognition technologies have led to wrongful arrests in other parts of Brazil. Since 2020, mistaken identifications have occurred in systems used in the Federal District, Bahia, and Rio de Janeiro. In one case, a black man in Bahia was wrongfully detained for 26 days, accused of a robbery that had been committed by someone else ten years earlier. Similarly, in January 2024, a woman in Rio de Janeiro was detained due to an error in the database: the system identified her as a fugitive, but she was already serving her sentence in an open regime.
Criticism
Public safety experts have expressed concerns about the risks associated with urban surveillance programs. Daniel Edler, a researcher at the University of Glasgow and the Center for Violence Studies at USP, warns that similar experiences worldwide have shown how these technologies can be misused and how uncertain their effectiveness is in combating crime. Edler cited a case in Rio de Janeiro where a police officer used the system for personal purposes: after a car accident, the officer accessed surveillance images to support a lawsuit and prove he was not at fault.
20,000 Cameras
The program currently has more than 90% of the 20,000 cameras contracted by the city government, and the company responsible has until April to deliver the others. Most of them are located in the city center and east side. In addition, 4,800 company cameras have already been integrated into the system – which corresponds to a little less than half of the goal. The city government still needs to advance in the integration of its system with the camera networks of public service providers, such as the Subway and CPTM.
Analysis:
The Smart Sampa program marks a significant step forward in leveraging surveillance technology to enhance public safety in São Paulo. With over 1,400 arrests and 28 missing people located in its first six months, the program demonstrates its potential to positively impact urban security, particularly in high-crime areas. However, the implementation of such technologies requires a delicate balance between enhancing safety and protecting individual rights.
The program’s early history raised legitimate concerns. The initial version of its public notice, which allowed for criteria like skin color and cases of vagrancy to identify suspects, highlighted serious ethical lapses in its planning phase. Although these issues were later addressed, they underscore the importance of rigorous ethical oversight in the deployment of such initiatives. Another critical issue is the reliability of the technology. While Smart Sampa claims to issue alerts only when there is a 90% confidence level, errors in similar systems across Brazil reveal that this assurance is not foolproof. Cases of wrongful arrests in Bahia and Rio de Janeiro, caused by flaws in databases or facial recognition, emphasize the need for robust auditing mechanisms and reparations for errors.
An effective system must not only apprehend criminals but also minimize mistakes and protect the rights and reputations of innocent individuals. Ultimately, the success of Smart Sampa depends not just on the technology itself but on the administration’s ability to monitor, refine, and improve the program, ensuring it serves as a tool for societal protection rather than a threat to individual freedoms.