Why we should oppose Automated Video Surveillance.

Automated Video Surveillance (AVS) has dug its roots and wired its plugs in France, with support from the central government, other local authorities and the French data protection authority, the CNIL (refer to our article “What is AVS?”). Our opposition is building up, both locally with the Technopolice campaign and at a national level with a response to the public consultation on AVS, recently organized by the CNIL. They are plenty of reasons to reject AVS, many of which we have gathered through your own contributions to us, 175 of which we then transmitted to the CNIL. As for now, here is the current state of our political motivations against AVS.

This article has been translated from it’s original publication in french by our community of volunteers. You can participate in this effort by following this link. Thanks !

As a key thread of our rationale, we defy our opponents’ reasoning that pretends to seek an equilibrium between a ‘security’ that AVS would provide and the measure of freedom we would need to sacrifice in return. To us, this dilemma is fallacious: AVS will not only prey on our freedom but also on our security. It will reduce the security of a large part of the population without countering the dangers it is claiming to tackle.

Negative effects on security

Threats raised by AVS over the security of a population are threefold: it endangers the populations that are already the most vulnerable ; on a systemic level, it favors violent behaviors by the police against the population ; it gives executive branches of the government such power that no counter-power will be able to prevent its abuse.

Putting the most vulnerable populations in danger

As with any public space surveillance system, AVS will primarily monitor those who spend the most time outdoors – people who, due to lack of resources, have little or no access to private places to socialize or live. In addition, AVS detects behaviors more effectively when it has been trained on a large number of image sequences picturing the same action. Thus, the most effectively detected behaviors will be those most often encountered on the street and in transportation – the behaviors typical of the populations that spend the most time there, regardless of whether these activities are legal or illegal.

It is precisely these behaviors that AVS providers highlight ¹For example, the RATP, Paris’ public transport operator, recently experimented in one the city’s main interchange metro station a system to track people who are static for more than 300 seconds. : panhandling, begging, static meetings. It is the lifestyle of precarious or working-class populations that will be targeted first, even though it hardly ever constitutes a crime. AVS will play the role of an automated ethnic profiling system based on social criteria, making it possible to multiply the number of sound alarms or human controls and to exclude a part of the population from the public space, further deteriorating their security – whether it is a question of degrading ²Exclusion through surveillance adds to the urban planning and urban development policies already deployed against precarious and working-class populations. their living environment or of distancing them from access to health care and other public services.

The focus of AVS on the poorest populations is not simply the ‘side effect’ of a recent technology that would still have some ‘biases’. Rather the opposite, AVS is precisely sold as a way to fight against behaviors defined as “abnormal”. Although they are perfectly common and “normal” for a large part of the population, branding these behaviors as the opposite allows to denigrate the populations that adopt them. Thus, AVS is as much a tool of social exclusion as it is a tool of political propaganda, the effect of which will be to instil the feeling that certain populations (arbitrarily chosen by VSA providers and their clients) are not “normal” and must be excluded from the public space.

Dehumanizing the population

AVS reinforces the distance between the police and the population. This distance is first of all physical: the interaction goes through screens and is carried out only in one direction. The distance is also intellectual: officers no longer have to understand, evaluate or anticipate the actions of other humans when a machine does it for them³Gregoire Chamayou. “A Theory of the Drone,” 2013. Specifically, the author discusses the loss of empathy caused by the distance between the drone pilot and his targets.. Dehumanized and taken away from any responsibility, the police is reduced to a tool for mechanical action on bodies. They are detached from empathy and consideration without which police violence can only explode. This same empathy without which even more people would have lost their lives in the face of the worst crimes committed by the police (as notably documented⁴In addition to various individual police initiatives during the occupation, the case of the failed deportation in Nancy illustrates how the empathy of a group of police officers saved hundreds of people. during the period of Nazi collaboration).

In a less apparent way, this technological distancing goes along with an austerity political agenda. Society is drying up its spendings on support and assistance to individuals, and is now only financing their disciplinary management. In a letter to the CNIL, the PACA region – France’s third most populated region – defended the experimentation of facial recognition in the vicinity of two high schools, claiming that this project was “a response to the growing gap between the need for security at school entrances and the human resources available in the schools, in the context of successive plans to reduce the number of public sector workers”. The supervisory staff, concerned and attentive, is replaced by machines whose only role is to open and close accesses. Or in Nîmes, where the city has taken almost 10 million euros from the water investment budget to spend it instead on the purchase of real-time an Automated Anomaly Detection software.

AVS deepens the dehumanization of social control, which was already a criticism addressed to regular video surveillance. This endless race is part of the general technological headlong rush that drives both the collapse of public services and the ongoing ecological disaster.

The population is also dehumanized: it is used as a guinea pig to train the algorithms. Not only do these algorithms see inhabitants of cities as a mass of data to be monetized for the account of morbid corporations,  populations also make the software more efficient despite them and thus enable to export it on the international market of surveillance. For example, the multinational company Idémia, refines its facial recognition applications in French airports with the PARAFE or MONA devices, and then sells facial recognition equipment to China and participates in mass surveillance and the Uyghur genocide.

Breaking down barriers against police abuse

Today, a large part of the police forces’ resources are focused on their most important and legitimate missions (crime, violence against people) due to the limited number of police officers. As a result, the police has limited time and resources to pursue activities that have little legitimacy (against vulnerable populations, against protesters) or that constitute abuses of power (repression of political opponents, persecution of minorities).

Tomorrow, AVS promises to erase this material limitation by increasing tenfold the operational capabilities of the police, enabling it to pursue missions of their choice, whether these missions are not legitimate or whether they constitute abuses. For example, while it is now extremely costly to detect every post signs criticizing the government during a protest, AVS ensures to eventually make this a triviality, facilitating on-the-spot arrests or, coupled with facial recognition, allowing mass prosecution of overly expressive opponents. Similarly, if the visual tracking of political opponents today implies such important human resources that these operations can only remain exceptional, AVS makes this easy by making it possible to track, at almost no cost, a person on all the cameras of one or several cities.

This change in scale is significantly transforming the way police power is enforced. From a specific action in response to “needs” that can be democratically debated, we are witnessing the emergence of an omniscient police force with the capacity to monitor and act on the entire population. With AVS, the current 250,000 police officers would see their authority level reach that of millions of agents not equipped with such technologies. This would be enough to reach the police/population ratio emblematic of police states.

This considerable multiplication of the police’s power will not be balanced by an equivalent multiplication of its counter-powers. As of today, installation of AVS equipments are taking place at a pace that is far too great for the CNIL or associations such as ours to be able to act against all of it in time and with sufficient detail. Tomorrow, the situation will be even more dramatic regarding the daily use of these systems: no authority, no judge, no parliament will be able to verify that each of the innumerable detections made every day does not contribute to an abuse of power. No one will be able to verify that AVS does not allow the police to illegally reduce security conditions of large parts of the population.

In addition to the risks of police abuse, this change of scale in the surveillance of public space contributes to the criminalization of a growing number of behaviors. Thus, for example, most VSA software seeks to detect littering, the failure to wear a mask, or people who are static in public space, without these developments having been democratically acted upon, resulting mainly from initiatives by private companies.

No positive effect on security

Dramatic degradations caused by AVS are not compensated by any advantage in terms of security. It is an inappropriate tool to fight against violence, both in terms of its object, the public space, and in terms of its functioning, the automation.

This double inadequacy is based on a distorted vision of the concept of “security” which, in AVS promoters’ discourses, is limited to a pure marketing argument disconnected from the way in which the population could concretely protect its physical and mental health, its living conditions, its housing and its means of self-fulfilment.

Inadequacy of monitoring

The object of AVS is public space. Yet, for the most part, violence against people does not take place in the public space. While sexual assaults almost always take place in a private context (91% are perpetrated by a person known to the victim), the vast majority of homicides, excluding conflicts between criminals, also takes place off the public space⁵See statistics for the Paris region between 2007 and 2013, chart 25..

This inadequacy between the object monitored and the purpose pursued is at the heart of numerous studies which, for a decade, have unanimously concluded that traditional video surveillance is ineffective, in particular, reports by the Cour des Comptes (France’s supreme audit institution), the LINC (France’s digital innovation laboratory) and other researchers.

This inadequacy is accentuated in the case of AVS, which, in order to function, must train on a large number of video sequences representing behaviors to be detected. The issue is, acts of violence against people are much less numerous in the public space than simple acts of degradation, foraying or begging. Consequently, the algorithm will have far fewer opportunities to train itself to detect acts of violence and will detect them much less effectively than other more anecdotal acts (the surveillance of which, as we have seen previously, will degrade the security conditions of the most vulnerable people).

Inadequacy of method

Prevention of violence against people is based on human and social work: personalized support, healthcare, field investigations, sociological analyses, reduction of inequalities or even simple field work. This human work has an unavoidable cost and is already largely under-funded, particularly in areas where poverty is the highest.

On the opposite, AVS, probably less expensive in the short term, is only able to detect certain offences (and among the less serious ones), without being able to treat root causes. A way to give the illusion of treating symptoms, without changing anything in the long term.

This is probably one of the rare advantages of AVS: offering a discourse that will create illusions in the short term for elected officials in need of an inspiring political project. This discourse is all the more seductive for elected officials that the AVS industry has been preparing for several years the right language elements and a confusing enough discourse to hope to deceive the public. Perfectly usual behaviors typical of less affluent populations are described as “abnormal“. “Security” is presented as an objective that has much more to do with the “cleanliness” of the city and the “security” of goods than with that of people. Is said to be “enhanced” or “intelligent” police surveillance that, on the contrary, will be “reduced” to purely automated tasks and stripped of all the empathy and consideration that make up human intelligence

In conclusion, contrary to what its promoters claim, AVS is a serious threat to our security. It will undermine the living conditions of a large part of the population, open up unprecedented political risks, and this without even succeeding in protecting us. In addition to being a serious threat to our security, AVS will at the same time sweep away our freedom of circulation, of assembly, of expressing our political opinions or of having a private life. We will come back to the details of infringement on freedoms caused by AVS in a future article that will include the legal analysis developed in our response to the public consultation organized by the CNIL.

↑1 For example, the RATP, Paris’ public transport operator, recently experimented in one the city’s main interchange metro station a system to track people who are static for more than 300 seconds.

↑2 Exclusion through surveillance adds to the urban planning and urban development policies already deployed against precarious and working-class populations.

↑3 Gregoire Chamayou. “A Theory of the Drone,” 2013. Specifically, the author discusses the loss of empathy caused by the distance between the drone pilot and his targets.

↑4 In addition to various individual police initiatives during the occupation, the case of the failed deportation in Nancy illustrates how the empathy of a group of police officers saved hundreds of people.

↑5 See statistics for the Paris region between 2007 and 2013, chart 25.