The ‘Ketchup Effect’

A closer look at the use of surveillance measures by public authorities in Sweden following 9/11 reveals an interesting duality.¹⁾ On the one hand, electronic surveillance mandates are extensive and still expanding. The threat of terrorism has served as a continuous shadow-presence in this context. On the other hand, 9/11 seems to have done little to immediately influence surveillance of the public sphere in Sweden. Instead, camera surveillance, in particular, came to be developed later and in response to a somewhat different set of concerns. Once it began, the development can perhaps best be described as displaying a ‘ketchup effect’; where you open the bottle and at first nothing comes out, and then it all comes out at once and you have effectively ruined your dish (which, depending on your view of ketchup, may have been doomed from the moment you picked up the bottle).

The acceleration in the development of public sphere surveillance in Sweden can be explained through the distinction between the acceptance of surveillance measures targeting ‘the other’, and the possibility to separate this ‘other’ from ‘ordinary citizens’, who should be left alone. This distinction is by no means unique to Sweden, but it has led to certain specific entanglements affecting public surveillance, the contexts where it has been deployed and when.

The impact of 9/11 and software pirates on the Swedish surveillance discourse

Since 1977, Sweden has had a specific act limiting the use of surveillance cameras. Motivated by privacy concerns, the act reflected a clear political view that widespread monitoring of the public was unacceptable. This view withstood reform, until 1998, when the first indication of a shift in this view became visible. Crime prevention became a priority, and rather than attempting to limit the use of surveillance cameras, the act instead introduced efforts to balance the positive outputs of surveillance with the need for privacy protections. Swedish surveillance scholar Fredrika Björklund has described that ‘monitoring changed from an extraordinary measure that lawmakers handled with a certain amount of reluctance to an ordinary measure integrated in the social control system’. While the reform can be seen as a paradigm shift in the Swedish approach to camera surveillance, it preceded 9/11 and has been framed by Björklund as result of a neo-liberal turn in policy, moving towards a more generalized suspicion.

One could expect that in the wake of 9/11, this preventive turn would be intensified, but this is not visible. When the public inquiry tasked with investigating the overall Swedish preparedness and counter-terrorism capacity following 9/11 presented its final report, public surveillance measures were largely absent. The inquiry instead focused on the potential need for increased mandates for electronic surveillance – including the need to consider preventive covert surveillance of communications. Similarly, a 2002 public inquiry report on camera surveillance highlighted its potential value for crime prevention. Neither mentioned terrorism nor framed the issue in a larger post-9/11 security context. Other legal reforms and inquiries in the terrorism context in the years following kept to this theme. However, the political justification of these reforms followed a common pattern; the measures targeted presumptive terrorists and criminals, and ‘ordinary people’ had no cause for concern. Consequently, the first preventive telecommunications surveillance act was enacted in 2007 with rather limited public attention. This relative disinterest changed in 2008, when the government proposed expanded mandates for the strategic signals intelligence performed by the Swedish Defense Radio Establishment (FRA). The scope of this surveillance would no longer be limited to radio communications, instead including communication crossing the Swedish border in fiberoptic cables – the first bulk-surveillance of telecommunications introduced in Sweden. In a development which took the government by surprise, surveillance became a topic of political import, leading to a minor parliamentary crisis and significant public pressure on the sitting government, despite efforts to play down the significance of the reform.

Of course, the general surveillance of telecommunications is worthy of political debate and judicial scrutiny, so the debate could be seen as a healthy and principled democratic response to a controversial bill. I would, however, like to advance two (rather more cynical) interrelated explanations which may have contributed to this sudden and rather uncharacteristic expression of privacy outrage in a country generally known for its high trust in public authorities. The first is that the issue coincided with an ongoing general debate on measures against illegal file-sharing, which had become something of a national past-time, and which would the following year propel the Swedish Pirate Party into the European Parliament. As such, increased scrutiny of internet traffic by public authorities likely raised somewhat bigger concerns among the general public than would otherwise have been the case. This also feeds into the second explanation. Namely, that for the first time, the proposed surveillance would actually carry a far-reaching potential of including ‘ordinary citizens’ in the group subject to surveillance. In other words, what used to be a (largely absent) discussion about the rights of ‘others’ – terrorist suspects – became a lively discussion about the rights of the common Swede. This potential of mass-surveillance to shift the paradigm of the debate, undermining citizen trust by turning them into ‘the other’, has been pointed out by Marie-Helen Maras.

Shaking the Ketchup bottle of public surveillance

While the proposed signals intelligence act was eventually pushed through, the associated debate may have generated some political restraint towards proposing measures of surveillance impacting the general public. Our short excursion into this debate may therefore perhaps begin to explain why, in the decade following the debate on the signals intelligence act, there were only minor expansions of public sphere surveillance. Some small scale pilot studies of public surveillance were initiated, showing rather unclear benefits in terms of crime prevention. Nevertheless, a reform of the legal framework on camera surveillance in 2012 placed further emphasis on crime prevention interests and eased restrictions on surveillance of stores, parking garages and the subway. These developments can be seen as precursors; the main government efforts to increase camera surveillance can instead be said to begin in 2018. By this time, several terrorist attacks had taken place that fueled the political incentives for further surveillance of public spaces: a failed terrorist attack against Christmas shoppers in central Stockholm in 2010, the devastating terrorist attacks in Oslo and Utøya, Norway, in 2011, and the terrorist attack using a delivery truck in central Stockholm in 2017, in addition to other major attacks in Europe. Still, when the needle began to move on the political interest in public surveillance, it was only partly on account of terrorism.

Since 2015, the Swedish police had begun to designate ‘vulnerable areas’; geographically delimited areas (primarily suburbs of larger cities) characterized by a low socio-economic status and where criminals have been deemed to have an impact on the local community. Targeting these areas for more intense policing increasingly became a political priority. Even more so, following a general increase in deadly violence in conflicts between criminal gangs, with vulnerable areas found to have eight times the amount of deadly violence in relation to its population size compared to the country at large. Police-operated camera surveillance in these vulnerable areas became a key tactic, along with community policing efforts and a general increase in the number of police officers on the ground. In a 2020 evaluation by the Swedish National Audit Office on the police initiatives in vulnerable areas, it was found that these areas contained over half of the 350 surveillance cameras deployed by the Swedish police. The expansion of surveillance cameras and the distribution of body-worn cameras to police officers remained as a priority.

This development was made possible through a rapid easing of the previously rather stringent requirements for deploying surveillance cameras in public areas. A reform in 2018 made it easier for the police authorities and municipalities to obtain permits for camera surveillance in public places for law enforcement or security purposes. The police authority and the Security Police would now also be allowed to use camera surveillance temporarily without a permit in the event of a risk of serious crime. The use of camera surveillance on metro trains and stations, as well as in health care facilities was also facilitated by easing the conditions for permits. Only a year later, the government fully removed the need for law enforcement agencies to seek permits to install new surveillance cameras in public. The recent exceptions made for metro transit were also significantly expanded to allow for surveillance of vehicles used in public transport in general, as well as stations, stops or areas associated with such transportation. Perhaps counter-intuitively, this removal of permit requirements was partly motivated by the increased protections for recorded data offered by the GDPR and the EU law enforcement directive. The associated requirements of internal proportionality assessments and controls were said to compensate for the removal of ex ante approval from an independent authority.

This ketchup effect of new surveillance opportunities strained the resources of the police. A research report from 2019 indicated a lack of manpower available to monitor and analyze surveillance video footage as a factor limiting its effectiveness. The growing availability of facial recognition tools is, however, likely to change this. Following a consultation with the Swedish data protection authority in 2019, the police has begun implementing facial recognition in its forensic analysis of video footage in ongoing investigations. Since then, legislative changes have enabled the further use of biometric data in testing environments ­– to develop further capabilities and enable the deployment of facial recognition in border controls. The increased interest in deploying surveillance cameras for law enforcement purposes has coincided with the emergence of tools to integrate these into police work in a cost-effective way. Of course, facial recognition tools change the privacy equation surrounding surveillance cameras. Privacy as a value in public is thus simultaneously circumscribed by the removal or easing of permit requirements, the emergence of privacy-invasive technologies ostensibly allowing identification of individuals captured in the surveillance footage, as well as the associated increased capacity to analyze larger amounts of footage through automation.

Nevertheless, there are still traces of restraint. The government has (so far) stopped short of proposing live facial recognition except as a development project in airport border crossing settings. The facial recognition implemented in the forensic analysis setting is not (yet) connected to existing police databases, but appears to be aimed at automated selection of video sequences where a specific likeness is visible. The political discourse is, however, clearly driven by a public security paradigm, fueled by the rise in gang conflicts and tied to previous legal developments against terrorism. Following the killing of a famous Swedish rapper (in a more affluent suburb of Stockholm) the leader of the opposition drew parallels between organized crime and terrorism. The government apparently agreed, as it has initiated a new public inquiry into the possible expansion of preventive surveillance measures developed for anti-terrorism purposes to the fight against organized crime.

The legal discussions regarding the expansion of facial recognition appears limited as well. In its analysis of the Swedish police authority’s proposed use of facial recognition comparisons, the Swedish Authority for Privacy Protection focused on the CJEU’s reasoning in Heinz Huber. The authority found that the requirement of necessity in the Law Enforcement Directive, as transposed in Swedish law, essentially asked whether the identification of suspects becomes more effective by the measure (than with manual controls). This interpretation places very few limits on the use of new investigatory measures or technologies. It also appears to extend the findings in Heinz Huber to a context not necessarily comparable, given that the case related to the previous data protection directive, and the increased awareness on the risks of facial recognition which has developed since then. The privacy authority has, however, been more assertive in relation to the unlawful use of Clearview AI by the Swedish police. Nevertheless, even there, there was no real constitutional debate on public surveillance. The judicial review of the Council on Legislation (Lagrådet) on the removal of permit requirements for law enforcement cameras focused on linguistic and minor legal-technical details, rather than any constitutional or rights-based concerns. Even the Swedish Authority for Privacy Protection’s predecessor offered primarily administrative and organizational comments on the government bill. This limited debate could itself signify a normalization of public surveillance.

Situating Swedish public surveillance in the discursive patterns of privacy and security

In ‘The Cultural Lives of Security and Rights’, Ian Loader points to the entanglement of security demands with “the production and reproduction of a ‘we’ whose territory, or values, or capacity for self-determination is felt to be under threat—either from without or from ‘enemies within’”. The current debate on deadly gang shootings in Sweden has, in many ways, released the Swedish debate on surveillance of the public sphere from a certain gordian knot of political sensitivity. Instead of legal reforms needing to account for the less palatable proposition of surveilling ‘the ordinary Swede’, the government can adopt a more convenient rhetoric where surveillance measures will primarily target ‘the other’. The definition of this ‘enemy’ also serves to distinguish this enemy from the public ‘we’. Even more so, as surveillance measures can be framed as targeting a group distinct from the general public, i.e., ‘criminal networks’, as well as the geographically and socio-economically distinct ‘vulnerable areas’ with a higher proportion of first or second generation immigrants.

The discursive patterns surrounding this emerging surveillance play into roles already established in the security paradigm that followed 9/11. The designated vulnerable areas subject to more intense surveillance also tend to be more segregated, which adds a further ‘otherness’ to their portrayal in political and media discourses. In designating the more severe level of ‘especially vulnerable areas’, the Swedish police authority takes into account, inter alia, the presence of “extremism, such as systematic violations of religious freedom or strong fundamentalist influence that restricts human rights and freedoms”, and “persons traveling to take part in combat in conflict areas”. It is perhaps here that we find the full implication of 9/11 on Swedish public surveillance – in the availability of discursive connections to terrorism ready for use and implementation into new preventive contexts. When the ‘the other’ who will be the primary target of surveillance can be credibly separated from the general public, the expansion of legal mandates of surveillance can follow. The fact that these legal mandates are not limited in scope to the areas or groups currently targeted is another matter. The proverbial ketchup has hit the plate and eventually everyone will get a taste of it.