No less than 90 Walloon and Brussels municipalities expressed their security concerns during the “48 hours of mayors”, addressing issues such as the fight against delinquency, crime and illegal deposits. The growing use of video surveillance, combined with the arrival of facial recognition, raises questions about privacy risks and the necessary legislative framework.
As the municipal elections approach, many candidates promise to increase the number of surveillance cameras in their municipality, with the aim of strengthening security, fighting against incivility, deterring troublemakers and finding the perpetrators. assaults. However, several municipalities have already taken similar measures.
For example, Tubize and Flémalle have installed cameras at strategic points to reduce illegal dumping, while Jurbise has invested in 121 cameras and reinforced the police presence. Uccle has installed 75 cameras and is working with citizens on prevention initiatives. La Bruyère and Berchem-Sainte-Agathe have also set up surveillance systems, reducing thefts by half in one and fighting against delinquency in the other.
Until now, surveillance cameras in Belgium do not use facial recognition, due to a lack of a specific legal framework authorizing its use. However, this could change with the upcoming entry into force of the AI Act, which aims to potentially allow this technology under certain strict conditions.
The AI Act: a European regulation dedicated to artificial intelligence
From February 2025, the European AI Act regulation, which governs the use of facial recognition based on artificial intelligence, will come into force. Its main objective is to prevent potential abuses linked to this technology and to guarantee supervised use. According to this regulation, the use of real-time facial recognition for law enforcement purposes is prohibited, except in three specific cases:
- The search for victims of kidnapping, disappearance or human trafficking
- Prevention of an imminent security threat or terrorist act
- Locating or identifying suspects involved in serious criminal offenses
However, this does not mean that this technology will necessarily be effective as of next February. “It’s even unlikely”warns Frank Schuermans, president of the body for monitoring police information. The entry into force of the European regulation cannot be achieved without specific legislation in each Member State. “In Belgium, this still requires a national legal framework which must be adopted to allow the use of facial recognition under the conditions provided for by the regulation”he explains.
According to him, there is therefore little chance that Belgium will have a law by this deadline. “States will have to determine which judicial or administrative authority will be responsible for issuing the necessary authorizations to use this system. It is also essential that the police and the government, if there is one, agree to make this possible from February 1 But there is little chance that Belgium will have adopted a law within a few months, because the development of such legislation is not simple and requires a serious and in-depth debate. “
On the issue of personal data protection, safeguards are provided for by the regulation, according to Frank Schuermans. “The approval of a judge or an independent administrative authority will be necessary. Each request must also be justified both in terms of necessity and proportionality.” It will therefore be necessary to prove that the same result cannot be obtained by other means.
In addition, the use of facial recognition must be limited in time and space, and “no legally binding decision can be taken solely on the basis of artificial intelligence.” In other words, police officers or magistrates will have to validate the results, which will have to be integrated into a broader body of evidence. “There are many guidelines to frame all of this”he assures.
A risk of discrimination for the Human Rights League
Although this technology is seen as a tool to protect the population, it raises many concerns, particularly from human rights defenders, who denounce the risks of abuse and excessive surveillance. Thus, the Human Rights League advocates an outright ban on facial recognition. “We are talking here about surveillance cameras, devices so common that they become almost invisible in our daily lives. From now on, these cameras will be able to identify us thanks to algorithmic software which compares the images captured in real time with data already existing”explains Emmanuelle de Buisseret Hardy, legal advisor to the Human Rights League.
This will increase repression and discrimination against these people since these databases are already initially set up with discriminatory biases.
The concern mainly concerns the transparency of police databases. According to Emmanuelle de Buisseret Hardy, there is in Belgium a “almost complete opacity” around police databases, in particular the BNG, a vast base which compiles information on both suspects and convicted persons. “In addition to this general database, the police can create other databases according to their desires and needs. The people most represented in these databases are those who already experience historical discrimination. So, it is these images which, first, will be accessible to the police. This will increase repression and discrimination against these people since these databases are already built up with human discriminatory biases. she explains.
Furthermore, access to personal data is “indirect and obstructed”because it must go through the police information control body, a process which, according to her, is not in compliance with European law. It also raises concerns related to cybersecurity: “No computer system is waterproof, and the risks of biometric data hacking are very real. Once this data, such as the unique characteristics of the face, disappears following a hack, a leak, or computer mishandling, it ‘is fucked up’she warns. She also underlines the risk of human excesses, affirming that the manipulation of data, entrusted to individuals, is “fallible.”
Finally, the legal advisor to the Human Rights League mentions a “slip” towards a standardization of surveillance, where the argument of security justifies the use of ever more invasive tools. “In fact, this is the foot in the door. Any first authorization will allow an increase in this list of exceptions.”
However, facial recognition has already been used by the Belgian police on several occasions. In 2017 and 2019, tests were carried out at Zaventem Airport by the Federal Police to evaluate the effectiveness of this technology. Additionally, in 2019 and 2020, police conducted approximately 70 searches using the controversial Clearview AI software. These experiments were, however, deemed illegal by the Police Information Control Body, because at the time there was no legal basis authorizing their use.
Finally, according to a study carried out in 2021 by KULeuven, among the 86 police zones in the Flemish and Brussels regions surveyed, at least 5 zones had facial recognition, one of them declaring to use it “often to very often”.
security facial recognition police privacy
Related News :