Algorithmic Governance, the Dutch Childcare Allowance Scandal, and the Importance of Legal Mobilization

1. Digitalization, AI, and datafication / 2. Algorithmic governance / 3. Human Rights and Data Justice / 4. Case-study: the Dutch SyRI system and the Toeslagenaffaire / 5. The Civic Coalition / 6. Concluding remarks

1. Digitalization, AI, and datafication

In the name of efficiency, efficacy, and social trust, governments and public administration increasingly take recourse to forms of digitalization and usage of algorithms in the implementation of policy. The usage of digital and algorithmic instruments can be related to different, in some ways overlapping tendencies. Such tendencies include: a. the increasing recourse to large datasets and big data, contributing to the quantification of governance; b. the increased usage of digital, non-human tools to exploit such data, in particular in the form of algorithms and the usage of AI and machine learning; c. an increased belief in the possibility of prevention by means of predictive instruments; d. the prioritization of efficiency over considerations of justice and individual rights.

In the contemporary world, with the advance of internet, social media, and additional digital tools, computerisation, digitalisation, and datafication are evermore ubiquitous phenomena (Diaz-Bone 2019: 77). Notions like ‘big data’, ‘data mining’, ‘algorithmic governance’ or ‘governance by numbers’, and ‘audit culture’ are frequently used to describe a world that is increasingly driven by numbers and data. In terms of public administration, this means an advance of ‘number-based governance’, where governments increasingly use and produce data sets, in order to ‘represent, analyze, and track their citizens’ (Diaz-Bone 2019: 78). Huge amounts of data are gathered by state organisations as well as by private or semi-private actors, which use this data for forms of control, surveillance, and risk prevention, or for making profit (Diaz-Bone 2019: 78). 

The combination of datafication and digitalisation means that possibilities for extensive surveillance by the state have grown enormously. In fact, Schuilenburg understands modern societies as ‘societies dominated by surveillance’ (2024: 36). The combination of digitalisation and AI can be understood as a ‘new infrastructure’ or ‘system technology’ with a ‘systemic effect for the whole of society’ (Schuilenburg 2024: 37). Digitalisation also means the rapid expansion of datafication, that is, the increased conversion of behaviour into data points (Schuilenburg 2024: 39). Through capillary digital means, the flow of digital data is continuous. Data is both gathered with specific purposes, but also through automated processes, and it is in particular the amassing of various datasets (gathered for specific, individual purposes) that creates potential violations of individual’s privacy. Contemporary societies can be understood as subject to ‘dataveillance’, where behaviour is continuously monitored, registered, and profiled, without individuals realising the process is going on (Schuilenburg 2024: 39). Individuals are put into different categories, based on their profiles, in a process of ‘social sorting’ (Schuilenburg 2024: 40). With the expansion of these new forms of data and algorithmic governance, the ‘[p]ublic recognition and usages of numbers as societal representations step by step became more and more detached from the societal control and knowledge about conditions of data production’ (Diaz-Bone 2019: 78). Individuals are not aware of their ‘data-double’, often do not know how and when data is gathered, which diminishes the standing of the citizen as an ‘autonomous and sovereign subject’ (Schuilenburg 2024: 40). 

Diaz-Bone, building on Foucault’s work on panopticism, has suggested the notion of ‘statistical panopticism’, indicating a ‘recent and powerful dispositive for the governance of the self and of others’ (Diaz-Bone 2019: 79). Datafication and quantification are now a predominant part in state governance processes, and allow authorities to profile and survey citizens without those citizens being aware of it. Statistical panopticism not only implies increasingly extensive surveillance of citizens regarding past behaviour (for instance, when applying for social subsidies), but it is increasingly geared towards prediction and prevention. This is particularly clear in AI-enabled forms of anticipatory governance, for instance in urban contexts, where real-time data interacts with AI data analysis and prediction (Cugurullo and Yu 2025). 

2. Algorithmic governance

Big data, algorithms, and machine learning are evermore at the heart of modern public administration and governance. Public governance is increasingly data-driven and based on the usage of algorithms, for the stated reasons of efficiency, efficacy (Duberry 2022), and fraud detection and prevention. The application of digital tools and AI is supposed to increase efficiency and effectiveness in governmental tasks, to improve the procedures and processes involved, and to increase the quality of public services’ (Duberry 2022: 15). The usage of big data, algorithms, and machine-learning allows for the detection of patterns and to automatically generate evaluations, predictions and decisions on their basis (Diaz-Bone and Schrör 2022: 25). The emphasis on efficacy and efficiency is, however, in important contrast to questions of social justice and ‘social negotiability’ (Issar and Aneesh 2021: 4).

Automation and digitalisation of the public administration is to some extent a process difficult to avoid in large-scale modern societies (Diaz-Bone and Schrör 2022). The problem, though, is when the usage of digital technologies develops in a way that ‘enables the bypassing of negotiation in any social process’ (Issar and Aneesh 2021: 5). In other words, the digitalisation of the system leads to forms of decision-making, surveillance, and monitoring that are not anymore directly under supervision of humans and that are opaque in terms of how decisions are actually made, and hence difficult to hold accountable. As Issar and Aneesh argue in their discussion of algorithmic governance, ‘algorithmic operations could reassemble and restructure power relations and institutions at the expense of the disenfranchised and impoverished’, that is, those that have least capacities to scrutinise and contrast the forms of social ordering by means of algorithms’ (Issar and Aneesh 2021: 6). 

Algorithmic governance tends to be opaque for technological and technocratic/expertise-related reasons. Many observers refer to algorithmic governance as suffering from ‘black box characteristics’, that is, ‘very few experts can understand how the most complex AI systems function, their lines of code evolve with the more data they are fed with … , and they are challenging to audit’ (Duberry 2022: 15). Processes of using algorithms (increasingly) develop an internal logic which is not fully open to the comprehension or scrutiny of human beings. But algorithmic governance also proves non-transparent and opaque due to explicit administrative decisions to keep the operation of algorithms hidden and use algorithms for purposes of surveillance and profiling (Issar and Aneesh 2021: 6-7).

It can be argued that algorithmic governance consists in a new mode of governance. Where traditional governance was based on either market-driven or authority/hierarchy-driven processes, algorithmic governance operates in a different manner. Admittedly, there is an affinity with hierarchical modes of governance, in terms of top-down rule-making and the enforcement of compliance (Kango 2025: 5). However, in contrast to hierarchical forms of governance, with ‘identifiable chains of authority’, ‘algorithmic hierarchies are often opaque and automated, making them less visible but highly pervasive’ (Kango 2025: 5). What is more, algorithmic governance displays a number of unique characteristics. First, its comprehensiveness, meaning that its comprehensively takes over entire tasks and uses vast amounts of data. Second, algorithmic governance works in an instantaneous fashion, allowing the rapid adjustment of rules, deliver feedback, and learn on the basis of continuous input. Third, algorithmic governance is not static but interacts with users and adjusts its functioning accordingly. And fourth, as already noted, algorithmic governance operates on the basis of invisibility, which means it becomes more difficult to uphold accountability and to contest policy decisions (Kango 2025: 6-7).

Issar and Aneesh identify three major areas of concern in algorithmic systems: surveillance, social bias, and individual identity (2021: 7). Algorithmic governance is in itself based on forms of quantification, which involves the simplification, classification, comparison and evaluation of complex social phenomena, and in relation to humans engages in forms of ‘profiling’ on the basis of disconnected data sets and points. The datafication of social phenomena and behaviour may lead to forms of dehumanisation. Surveillance involves forms of loss of privacy, but it also sees more intense forms of ‘asymmetrical dominance’, as those being under surveillance may not be aware of the fact (2021: 7).

Algorithmic governance may involve serious forms of social bias, in that existing forms of discrimination and social inequalities may become amplified. The bias may stem from problematic, biased input (‘garbage in, garbage out’, Issar and Aneesh 2021: 8), but can also be the outcome of the programmers’ values and principles in the coding process, and the way in which quantification and categorisation of people, objects, and events is programmed (Diaz-Bone and Schrör 2022: 26). In fact, algorithms are not neutral, contextless dispotifs, but are crafted by humans. As such, algorithmic governance ‘can be conceived of as the power to classify and quantify persons, objects and events on the basis of social conventions how to categorize and to quantify’ (Diaz-Bone and Schrör 2022: 26). Algorithmic governance may hence contribute to make forms of inequality and discrimination even more engrained through automated processes of selection and decision-making.

Algorithmic governance further has major implications for individual identity and self-identification. As Issar and Aneesh argue, big data and machine-learning through algorithms collect data without a specific purpose, leading to the profiling and identification of patterns of behaviour unrelated to the actual identities of individual persons. People hence become related to separate ‘system identities’ of which they are not aware (Issar and Aneesh 2021: 9) which may be in stark contrast to people’s self-perception. This may lead to arbitrary identifications in algorithmic systems which focus on patterns of relationships but not on actual people and their identities (Issar and Aneesh 2021: 9).

3. Human Rights and Data Justice

Algorithmic governance tends to have major implications for a number of core human rights. As indicated by the Council of Europe, there are important reservations as to the opacity and unpredictable nature of algorithms as used in automated governance processes (CoE 2018: 10). Beyond such general concerns, and in relation to the specific impact of algorithmic governance, according to the CoE the following human rights areas are effected by the use of algorithms: fair trial and due process; privacy and data protection; freedom of expression; freedom of assembly and association; effective remedy; prohibition of discrimination; social rights and access to public services; and the right to free elections (CoE 2018). The CoE recognizes that this is not an exhaustive list, and other forms of impact are possible, in relation to practically any human right, including the right to life (CoE 2018: 32-3). 

In the more circumscribed area of algorithmic governance (rather than the usage of algorithms per se), one can identify a number of prevalent human rights concerns. The first, most significant rights area regards the right to privacy and data protection. As stated in article 8 - Right to respect for private and family life - of the ECHR: 

1. Everyone has the right to respect for his private and family life, his home and his correspondence. 

2. There shall be no interference by a public authority with the exercise of this right except such as is in accordance with the law and is necessary in a democratic society in the interests of national security, public safety or the economic well-being of the country, for the prevention of disorder or crime, for the protection of health or morals, or for the protection of the rights and freedoms of others.

As indicated in art. 8(2) ECHR, the interference with the exercise of the right to privacy can only be justified in specific cases, and such interference needs to be proportional and balanced. The right to privacy is crucial in relation to algorithmic governance and Big Data (Bekker 2020: 297) as data-oriented governance utilizes existing, separate datasets and links these datasets together, leading to potential problems of proportionality and potentially arbitrary forms of governance actions. As article 8(2) ECHR states, interference with privacy by public authorities needs to be strictly within the limits of the law and can - inter alia - only be justified in the name of national security, public safety, disorder and crime, or the protection of the rights and freedoms of others. Large-scale data matching, or the linkage of personal data that has been gathered by various public authorities can hence only be justified if there are clear and pressing social needs to do so (Bekker 2020: 298). The collection of personal data for specific purposes may be legitimate, but the matching of various data-sets facilitates more or less extensive forms of profiling that may go much beyond the original purpose of the single data collection. 

Concerns such as transparency, the availability of information, the explainability of how and when algorithms are used, the construction and usage of risk indicators, and the linkage of siloed datasets are of crucial importance (cf. Rachovitsa and Johann 2022: 2-3). The right to privacy hence goes hand in hand with citizens’ right to information and transparency in the application of methods of governance. Citizens should be able to know how and when their data is being used, and for what specific purposes. 

A second area of major concern is that of non-discrimination, on the basis of for instance social origin, property or other status, and in relation to the right to privacy as well as the right to social security. ‘Predictive analytics, algorithms and other forms of AI are highly likely to reproduce and exacerbate biases reflected in existing data and policies’ (Rachovitsa and Johann 2022: 12). Research indicates that digitalisation and algorithmic governance may have discriminatory effects on specific groups, in particular vulnerable communities (Rachovitsa and Johann 2022: 12). According to Rachovitsa and Johann, however, the ‘difficulties of substantiating the (potential) discriminatory effect of algorithmic systems cast doubt on whether data protection rules and international human rights, as they currently stand, are well-suited to address risks posed to specific groups’ (2022: 13). 

A third highly significant area is the right to remedy and access to justice. According to the CoE, states ought to ‘ensure that individuals have access to judicial or other procedures that can impartially decide on their claims concerning violations of human rights online, including effective non-judicial mechanisms, administrative or other means for seeking remedy such as through national human rights institutions’ (2018: 23). Such mechanisms include effective complaint mechanisms, but also transparency and access to information with regard to specific policy practices. In particular in the context of algorithmic governance, the right to remedy and access to justice may be negatively affected: ‘The nature of decisions being made automatic, without or with little human input, and with a primacy placed on efficiency rather than human-contextual thinking, means that there is an even larger burden on the organisations employing such systems to provide affected individuals with a way to obtain remedy’ (CoE 2018: 24).

4. Case-study: the Dutch SyRI system and the Toeslagenaffaire

The usage of algorithms, AI, large datasets, and forms of risk prediction in public policy, by governments and public authorities, is increasing exponentially. An important area where digitalisation is prominent is in the context of the welfare state and social services, and the provision of welfare and social subsidies. In the name of fairness, efficiency, and efficacy, public policies are driven by digitalisation, large datasets, and the usage of algorithms and AI to ‘automate, predict, surveil, detect, target, and punish’ (UN 2019, in: Bekker 2020: 290). Such policies, however, heighten the risk of producing policy outcomes that are in tension with human rights. 

The case-study discussed here is an important example of large-scale human rights violation – with major political[1] and juridical consequences – due to public policy based on algorithmic governance, with as prominent dimension a court-ruling identifying important violations of ECHR rights[2]. The case at hand is the usage of the digital system called System Risk Indication (SyRI) by the Dutch fiscal authorities, in particular in the specific context of social security fraud. SyRI was created with a specific purpose in mind: to focus on and detect fraud in vulnerable neighbourhoods in the Netherlands. Under the SyRI system, Dutch citizens were surveiled without being aware of it, or without knowing what kind of behaviour might trigger an official suspicion of engaging in fraud (Bekker 2020: 292). 

Here, we will look in particularly at the application of SyRI in the context of state subsidies for parents’ use of childcare. This led to the Toeslagenaffaire or Dutch Childcare Allowance Scandal (Hadwick and Lan 2021). The Childcare Allowance Scandal involved more than 35.000 parents. According to Hadwick and Lan, the Scandal is a ‘textbook example of failed algorithmic governance’ and ‘unprecedented injustice’ (2021: 610). One of the important triggers of the rigid and invasive policy regarding childcare benefits was a media storm around the so-called Fraud of Bulgarians[3] in 2013. Main victims of the Childcare Allowance Scandal were the parents that faced recovery claims by the fiscal authorities of large sums of received benefits. The Dutch childcare system pays an allowance in advance, with the final amount calculated at the end of the year. Moreover, the amount of benefits fluctuates on the basis of the changes in income of the received (Hadwick and Lan 2021: 612). In order to prevent fraud, the Dutch fiscal authorities developed an ‘all-or nothing’ approach, meaning that rather than only reclaiming a part of the benefits in case for instance irregularities were found, the entire sum (an entire year of benefits) was reclaimed in cases of suspected fraud (Hadwick and Lan 2021: 613). 

A core dimension of the entire policy was the usage of algorithms in auditing childcare benefit-receiving parents. A machine-learning ‘risk management system’ (RMS) was applied to preselect parents for audits. The system would select specific agencies with high levels of fraud, often in disadvantaged neighbourhoods, where often people with a migration background live. On the basis of a relatively small number of indicators, such as the distance of domicile to the child care facility or the foreign background of recipients, parents were selected for auditing (Hadwick and Lan 2021: 621). 

The Childcare Allowance Scandal involves various tensions with the law, and infringements of human rights. A first aspect is the non-proportionality in the implementation of the childcare policies. As mentioned, this in practice may have meant that purely due to a clerical or bureaucratic error, a recipient would have had to repay an entire, annual sum of benefits, which could run into tens of thousands of euros (Hadwick and Lan 2021: 612). A second aspect is the lack of access to information and justice. It can be argued that articles 6 and 13 of the European Convention of Human Rights are of relevance, in terms of the right to fair trial and the right to an effective remedy (Hadwick and Lan 2021: 615). The Dutch fiscal authorities engaged in a number of violations, that is, using a group rather than individual approach (e.g. through ethnic profiling), not providing reasoned decisions to recipients, and not granting the right to be heard, making it impossible to object to decisions (Hadwick and Lan 2021: 616, 618). A further infringement regards the right to privacy (art. 8 ECHR) and data protection of welfare recipients. This is directly relevant for algorithmic governance, as recipients had no means to access the data retained by the administration, including on how individuals have been profiled. Moreover, the way the algorithms operate, on the basis of what kind of initial input as well as forms of machine learning while operating makes it very difficult to hold the authorities accountable (Hadwick and Lan 2021: 625). 

In 2018, five Dutch human rights and civil society organisations, as well as two public persons (journalists and intellectuals), together with a major trade union, started a litigation procedure against the Dutch state at the District Court of the Hague, on the basis of claims regarding various violations of human rights as enshrined in the ECHR, the European Charter of Fundamental Rights (ECFR), and the General Data Protection Regulation (GDPR)[4]. On 5 February 202, the court issued its judgment int the case NJCM et al. De Staat der Nederlanden. The court argued that the Dutch state failed to ‘strike the right balance between the benefits the use of such technologies [algorithms, SyRI] brings as regards preventing and combating fraud’ and ‘the ‘potential interference with the exercise of the right to respect private life through such use’ (ECLI 2020). The court argued that ‘[u]nder the legislation all persons should be able to have the reasonable expectation that their private life is sufficiently respected when SyRI is applied. The court holds that the SyRI legislation does not meet this requirement’. In fact, the ‘SyRI legislation does not meet the requirement laid down in Article 8 paragraph 2 ECHR that interference with the exercise of the right to respect for private life is necessary, proportionate and subsidiary in relation to the intended purpose’ (ECLI 2020). The court continued: 

The court is of the opinion that the legislation does not strike the ‘fair balance’ required under the ECHR between the social interest the legislation serves and the violation of private life to which the legislation gives rise to qualify as a sufficiently justifiable violation of private life. The court takes into consideration the fundamental principles underlying the protection of data under Union law (the Charter and the GDPR), specifically the principles of transparency, purpose limitation, and data minimisation (ECLI 2020).

5. The Civic Coalition

The judgment of the District Court of The Hague was the outcome of longer-term civil society action and shows the great importance of forms of societal political and legal mobilisation to publicize and contrast negative outcomes of algorithmic governance. As stated in a recent research project on algorithms and algorithmic governance – “The Human Error Project” - of the University of St. Gallen: ‘[o]ver the last five years we witnessed the emergence of multiple debates about the human rights implications of AI and algorithmic profiling. These debates were often initiated and reinforced by the work of civil society organizations, who focused their energies and actions on raising awareness towards algorithmic injustice amongst the public opinion, policy makers and the media’ (Scharenberg, Barassi and Di Salvo 2024: 5)[5]. Civil society organizations act as civil watchdogs or “counter-democratic” forces, scrutinizing and monitoring government policies in relation to surveillance, profiling, risk prediction, and automated decision-making.

In the case of the Dutch Childcare Allowance Scandal, a Dutch civic coalition was formed out of the mobilisation of several civil society organisations[6] in 2016 in order to hold authorities accountable and inform the wider public about the negative implications and human rights consequences of algorithmic governance. This coalition started with the filing of a Freedom of Information request (FOI) with the minister of Social Affairs and Employment in 2017, to ultimately move towards legal mobilisation and strategic litigation in 2018.

Legal mobilisation is a phenomenon that is increasing in importance in European countries as well as on the European level. Legal mobilisation includes ‘strategic litigation’ or the ‘use of legal action to bring about certain social, political or legal changes’ (Handmaker 2026: 5). It needs to be recognized that the strictly legal dimension of litigation is often accompanied by other forms of action, directed at informing and involving the public as well as seeking political impact. As stated above, in the context of algorithmic governance, civil society-driven political and legal mobilisation are of great importance for human rights protection. In earlier work, I have proposed that the legal mobilization of social movements may be broken down in (a combination of) the following actions (Blokker 2012; 2020): 

a. Litigation in courts. Individual and collective actors engage in judicial and/or administrative proceedings to pursue objectives in relation to human rights and legal claims; 

b. Advocacy campaigns, lobbying, and engaging political institutions. The legal strategies of collective societal actors include the interaction and engagement with public institutions in order to promote political-legal objectives (such as human rights protection);

c. Mobilization and claims-making. The making of political and legal claims may involve the organization of public gatherings (deliberative fora), public claims/protests (in the name of human rights protection), or the making of comprehensive legal/accountability claims in public debate, as well as references to international arrangements/conventions or other legal and constitutional contexts. The addressees of claims include institutions as well as the wider public;

This is not an exhaustive list, and these different forms of action are frequently utilized in a parallel or sequenced manner. 

In line with what is put forward in the literature on political and legal mobilisation (e.g. Anagnostou 2014; Boutcher and McCammon 2018; Bürli 2018; Cichowski 2007; Conant et al. 2018; De Búrca et al. 2022; Handmaker 2026), the goals of the Dutch civic coalition, which included an organisation of lawyers, were at least two-fold (Wieringa 2023: 14). First, the ‘tactical’ or ‘public’ dimension of mobilization aimed at informing the public and mobilise citizens’ engagement in defence of individual rights. This approach included a request of Freedom of Information in 2017. It also included the inclusion of two well-known journalists in the coalition, who became public advocates of the coalition’s cause, participating in well-known television shows and engaged extensively in other media (Wieringa 2023: 16). The public campaign further included a media campaign, organised by the Platform Bescherming Burgerrechten, entitled ‘Presumed Guilty’ (‘Bij Voorbaat Verdacht’)[7]. The campaign sought to mobilise a public discussion on ‘risk profiling’ in relation to citizens’ rights and freedoms, the stimulation of research and analysis of the effects of algorithmic governance to enhance transparency, to educate citizens on their rights, and to empower citizens in their rights’ protection[8].

Second, the coalition developed a more ‘strategic’ set of actions, which included advocacy and legal mobilization in the stricter sense of engaging in legal action or strategic litigation. This resulted in the court case at the District Court of The Hague, mentioned above, in which the coalition criticized the Dutch state for fundamental rights violations, forms of discrimination against specific groups, and in general, and the attempt to create a ‘surveillance society’, resulting from the usage of the SyRI system[9]. In the court case, the United Nations Special Rapporteur on extreme poverty and human rights Philip Alston also contributed by sending an amicus curiae intervention. The SyRI case was of particular interest to the Special Rapporteur for a number of reasons.

First, this is one of the first legal challenges that the Special Rapporteur is aware of that fundamentally and comprehensively contests the systematic, legislatively sanctioned, use of digital technologies in the welfare state on human rights grounds.10 Second, this is the first example of litigation that the Special Rapporteur has come across in which the use of digital tools to prevent and detect welfare fraud has been challenged on human rights grounds. Third, the Special Rapporteur has serious concerns about the specific targeting of poor and other vulnerable groups by the Dutch authorities through the SyRI system and the projects that preceded it. It appears that SyRI is predominantly used in areas where there are higher concentrations of such poorer and vulnerable groups… (Alston 2019: 2-3).

The coalition engaged in litigation not so much to defend specific individuals and a specific verdict but rather sought to stimulate public debate on the future of algorithmic governance and the risk of a ‘surveillance society’ (Wieringa 2023: 15). As observed by Iwan-Sojka, the ‘procedure did not aim at paying compensation to victims but rather at removing the SyRI’s deployment from the legal procedures’ (Iwan-Sojka 2025: 105). In general, he argues, the ‘engagement of NGOs in public interest litigation was focussed on annulling the risk model rather than seeking justice for individual victims’ (2025: 105). In an interview in 2018 (Van der Sloot and Van Schendel 2019: 216ff), one of the partners in the Dutch civic coalition, Privacy First, stated in fact that its main aim is promoting the common good (rather than individual victims or group interests). While Privacy First engages in strategic litigation, this is normally the final step in a broader and multi-faceted strategy. The representative of Privacy First argued that the first step is gathering and exchanging information and knowledge on specific themes, which in the case of Privacy First is evidently the right to privacy. In a second step, Privacy First engages in lobbying and ‘silent diplomacy’ to generate attention for specific privacy issues, to, in a third step, engage in a public campaign, involving media, open letters to Parliament, the publication of statements on its website, and the mobilisation of the organisation’s supporters. Only when the three steps do not have the desired effect, the organisation may start up a process of strategic litigation. The importance of a judicial case for Privacy First lies in: giving a strong signal to the legislature, inform and mobilise public opinion, and stimulating the development of jurisprudence on the particular case, and possibly hoping for setting a juridical precedent (Van der Sloot and Van Schendel 2019: 216). 

In the Childcare Allowance Scandal, the civil coalition played a highly significant societal role in promoting civil awareness and informing the public, putting pressure on authorities, and endorsing the protection of human rights through litigation. According to the civil society organization Algorithm Watch, ‘Dutch activists got an invasive fraud detection algorithm banned’[10]. According to Krommendijk and de Vries, the role of societal organizations in strategic litigation and human rights protection in the Netherlands as well as on the European level has in recent years led to ‘important and in some cases groundbreaking judgments’, including in the SyRI case (Krommendijk and de Vries 2025: 2).

6. Concluding remarks

Algorithmic governance is advancing in modern democracies and becoming increasingly part of various policy domains. It is clear that the use of algorithms, machine learning and AI is frequently in contrast with fundamental human rights, including the right to privacy, to non-discrimination, and access to a remedy. While in the Dutch case of the application of SyRI in the context of social security policies Dutch authorities have at least partially been held accountable, not least significant civil society action in the form of legal mobilization and public interest litigation, this unfortunately does not mean that the tensions between algorithmic governance and human rights have attenuated. In fact, in various ways, the Dutch government has engaged in new policies that may be even more invasive regarding individuals’ rights[11]. One policy area where similar practices as with the Childcare Allowance Scandal have been found is that of student loans, where in particular students with a foreign background were singled out by Dutch authorities for auditing purposes[12]. Civil society, together with investigate journalism, and rights’ monitoring authorities, are playing a highly important role in enhancing public awareness on the dangers of algorithmic governance, and in contrasting forms of ‘surveillance society’ by means of legal mobilization. What is really needed, however, is a drastic change in the relation between public governance and the use of algorithms and AI, where the latter is duly covered in a set of clear norms and rules. In current circumstances, AI-driven anomie is only set to increase, exposing citizens to potentially extensive rights’ infringements and extensive forms of surveillance.

Bibliography

Algorithm Watch (2020), ‘How Dutch activists got an invasive fraud detection algorithm banned’, available at: https://algorithmwatch.org/en/syri-netherlands-algorithm/, accessed on 15 March, 2026.

Alston, P. (2019), Brief by the United Nations Special Rapporteur on extreme poverty and human rights as Amicus Curiae in the case of NJCM c.s./De Staat der Nederlanden (SyRI) before the District Court ofThe Hague (case number: C/09/550982/ HA ZA 18/388), available at: https://www.ohchr.org/sites/default/files/Documents/Issues/Poverty/Amicusfinalversionsigned.pdf, accessed on 15 March, 2026.

Anagnostou, D. (2014). Rights and Courts in Pursuit of Social Change. Hart Publishing.

Bekker, S. (2020). Fundamental rights in digital welfare states: The case of SyRI in the Netherlands. In Netherlands yearbook of international law 2019: Yearbooks in international law: History, function and future (pp. 289-307). The Hague: TMC Asser Press.

Blokker, P. (2012). Constitutions and democracy in post-national times: a political-sociological approach. Irish Journal of Sociology, 20(2), 68-90.

Blokker, P. (2020). Constitutional resistance in populist times. Federal law review, 48(4), 511-528.

Boutcher, S. A., and H.J. McCammon (2018). ‘Social movements and litigation’. The Wiley Blackwell companion to social movements, Wiley Blackwell, pp. 306-321.

Bürli, N. (2018), Third-Party Interventions before the European Court of Human Rights, Cambridge/Antwerp/Portland, Intersentia.

Cichowski, R. A. (2007). The European court and civil society: Litigation, mobilization and governance. Cambridge University Press.

Conant, L., Hofmann, A., Soennecken, D., and L. Vanhala, L. (2018). ‘Mobilizing European Law’. Journal of European Public Policy, 25(9), pp. 1376-1389.

Council of Europe (2018), Algorithms and Human Rights. Study on the human rights dimensions of automated data processing techniques and possible regulatory implications, DGI(2017)12.

Cugurullo, F., & Xu, Y. (2025). When AIs become oracles: generative artificial intelligence, anticipatory urban governance, and the future of cities. Policy and Society, 44(1), 98-115.

De Búrca, G. (ed.). (2022). Legal mobilization for human rights. Oxford University Press.

Diaz-Bone, R. (2019). Statistical panopticism and its critique. Historical Social Research/Historische Sozialforschung, 44(2 (168), 77-102.

Diaz-Bone, R., & Schrör, S. (2022). Algorithms, conventions and new regulation processes. In Democratic Frontiers (pp. 24-46). Routledge.

District Court of the Hague (2020), Judgment of 29 January 2020 in the case C/09/554385 / HA ZA 18/647, ECLI:NL:RBDHA:2020:865, available at: uitspraken.rechtspraak.nl/inziendocument?id=ECLI:NL:RBDHA:2020:1878, accessed 15 March, 2026.

Duberry, J. (2022). Artificial intelligence and democracy: risks and promises of AI-mediated citizen–government relations. In Artificial intelligence and democracy. Edward Elgar Publishing.

Hadwick, D., & Lan, S. (2021). Lessons to be learned from the dutch childcare allowance scandal: a comparative review of algorithmic governance by tax administrations in the Netherlands, France and Germany. World Tax J., 609.

Handmaker, J. (2026). Legal Mobilization: Assessing the Potential and Limits of Law-based Civic Advocacy. In Legal Mobilization. Edward Elgar Publishing.

Issar, S., & Aneesh, A. (2022). ‘What is algorithmic governance?’. Sociology Compass, 16(1), e12955.

Iwan-Sojka, D. (2025), Legal standing in algorithmic harm cases. Collective action in litigating risk models across e-government states, Studia Iuridica, (106), 94-113.

Kango, U. (2025). Algorithmic governance. In Handbook of Human-Centered Artificial Intelligence (pp. 1-26). Singapore: Springer Nature Singapore.

Rachovitsa, A., & Johann, N. (2022). The human rights implications of the use of AI in the digital welfare state: Lessons learned from the Dutch SyRI case. Human Rights Law Review, 22(2), ngac010.

Schuilenburg, M. (2024). Making surveillance public: Why you should be more woke about AI and algorithms, Den Haag: Boom Uitgevers.

Van der Sloot, B., & Van Schendel, S. (2019). De modernisering van het Nederlands procesrecht in het licht van big data. Procedurele waarborgen en een goede toegang tot het recht als randvoorwaarden voor een data-gedreven samenleving. TILT/WODC, Tilburg.

Van Stokkom, B.A.M. (2023), ‘ToeslagenGate. Hoe het geloof in bedrijfsmatig werken en de obsessie met fraudebestrijding de overheid corrumpeerde’, Letter to editor, Tijdschrift over cultuur en criminaliteit, 13, 1, (2023), pp. 77-92.

Wieringa, M. (2023). “Hey SyRI, tell me about algorithmic accountability”: Lessons from a landmark case. Data & Policy, 5, e2.

1. In January 2021, the entire cabinet of prime minister Mark Rutte resigned due to the publication of a report of parliamentary investigation on the Childcare policies, entitled ‘Unprecedented Injustice’ (Ongekend Onrecht) (Hadwick and Lan 2021: 610).

2. District Court of The Hague, SyRI (System Risk Indication), ECLI:NL:RBDHA:2020:187 (5 February 2020).

3. This fraud consisted in organised crime using the Dutch system to obtain childcare benefits which were obtained under false claims and then benefits channelled to Bulgaria (Van Stokkom 2023; Hadwick and Lan 2021: 621).

4. See ‘Dagvaarding’, https://bijvoorbaatverdacht.nl/wp-content/uploads/2019/04/dagvaarding-bodemprocedure-syri-27-maart-2018.pdf. Accessed on 14 March, 2026.

5. One notable initiative is the setting up of “Algo-Rules” by a number of experts, and independent think-tanks, including the Bertelmannsstiftung and iRights.Lab. The Algo-Rules ‘focus on those algorithmic systems that have a significant impact on society or individual lives, regardless of whether this involves direct or indirect effects. It is therefore not a question of establishing a set of design rules for all algorithmic systems but, rather, for those that are socially relevant. In order to determine whether an algorithmic system is socially relevant, an impact assessment should be conducted before it is designed. The stronger the potential influence of an algorithmic system on society or people’s lives, the more carefully it should comply with the Algo.Rules’, available on https://algorules.org/fileadmin/files/alg/Algo.Rules_EN_2.pdf, accessed on 20 January 2022.

6. The organisations are: Stichting Platform Burgerrechten, Nederlands Juristencomité voor de Mensenrechten, Stichting Privacy First, Stichting KDVP, De Landelijke Clientenraad, and FNV. In addition, two individuals, journalists and writers, joined the civil coalition. A number of these organisations set up the Privacy Coalitie in 2022.

7. See https://bijvoorbaatverdacht.nl/, accessed 15 March, 2026.

8. See https://bijvoorbaatverdacht.nl/missie/, accessed 15 March, 2026.

9. See the coalition’s plea, available at: https://bijvoorbaatverdacht.nl/wp-content/uploads/2019/10/20191029-Pleitnotities-NJCM-c.s.-inzake-SyRI.pdf. Accessed on 15 March, 2026.

10. See https://algorithmwatch.org/en/syri-netherlands-algorithm/, accessed on 16 March, 2026.

11. See, for instance, https://www.binnenlandsbestuur.nl/digitaal/verkiezingen/burgercoalitie-wil-dat-eerste-kamer-super-syri-afwijst; https://www.ftm.nl/artikelen/wet-gegevensverwerking-schaadt-privacy, accessed on 16 March, 2026.

12. See https://www.ftm.nl/artikelen/belastingdienst-weigert-openheid-algoritmen?share=pYRMN8hd1ATPx%2BDMuVpuLd38H0t9Xbs%2F4W3NLGCTYGUj5mSUQbCe89D5M0w9UkQ%3D, accessed on 16 March, 2026.