Principles for the Digital Society and the use of AI technology – Implications for public safety practice

Blogpost for the AI-MAPS project
HSD Campus workshop 8 February 2024

Blogpost by Sylvia I. Bergh, Merel Heijke, Mark Ruijsendaal, Tim Kies, and Nanou van Iersel.

Earlier this month, AI-MAPS researchers participated in a workshop at the Security Delta Campus, co-organized with the ICSS, HSD and VNG, on the Dutch ‘Principles for the Digital Society’ and what they mean for professional practices at municipalities and technology companies. As researchers, we were of course also very keen to collect relevant research questions for the AI-MAPS project’s second use case on fixed private cameras (such as ‘smart’ doorbells or shop cameras), which sometimes also record public spaces.

At the end of 2022, the Annual General Meeting of the Association of Dutch Municipalities (VNG) unanimously approved the Principles for the Digital Society, which are meant to guide municipalities in policy and practice regarding for example digital data collection and the use of biometric surveillance technology (AI-MAPS researchers contributed a position paper during the consultation phase in 2022). Municipalities implement and embed the principles in their purchasing conditions and procurement guidelines at their own pace, but not much is known yet about whether and how the principles are workable in practice.

The workshop aimed to address the following questions: How can municipalities, but also technology suppliers, apply the principles in their practice, e.g. in procurement and design requirements? What does this potentially mean for products and services offered by the private sector? How workable and understandable are the Principles? And what about upcoming European legislation? 

According to Tim Kies from the VNG, Dutch municipalities are important players when it comes to buying ICT solutions – they spend 1,8 billion Euros annually on them. So they can use this leverage to enforce some conditions on the use of this public money in order to protect public values such as transparency, democratic decision-making procedures, and privacy. These values are in line with the EU’s body of 13 laws under the heading ‘A Europe Fit for the Digital Age’ and the declaration on European Digital Rights and Principles

With regard to self-learning (AI) technology, principle 3.5 for example imposes a heavier proportionality-test. Principle 3.7 is also relevant as it states that every dataset, sensor and algorithm has a civil servant assigned to it who is known and accessible and can be held responsible for its use. Similarly, principle 4.2 prescribes that algorithms and sensors that are used in public space are registered in an algorithm-  and sensor registry.

HSD Campus workshop 8 February 2024 compilation picture

A lively discussion took place among the 30+ participants from security technology companies, the police, and those charged with public safety in municipalities. The algorithm register of the municipality of Amsterdam was discussed – for example, who should be responsible for registering a product – the municipality or the producers? And can the registration be done more efficiently if for example two municipalities use the same technology with the same algorithm? Another hotly debated issue was the municipality of Utrecht’s policy to impose a maximum limit on the number of cameras in public space. An overarching challenge is to get the principles embedded in an organisation and its processes, and understood and applied by both civil servants, mayor and aldermen, and the municipal council.

The discussion linked very well to the short presentation on the AI-MAPS project and its second use case, given by Sylvia I. Bergh. For example, principle 6.5 states that ‘when a private party or individual citizen places sensors in the public space, that party is bound by all applicable laws and regulations as well as additional frameworks and guidelines.’ An emerging coalition called ‘The Friendly Doorbell’ (Coalitie de Vriendelijke Deurbel in Dutch) is now working on developing such additional guidelines. These are badly needed, given that there are already 1.2 million of such video-doorbells which lead to an avalanche of complaints to the Dutch Data Protection Agency (Autoriteit Persoonsgegevens) and court cases between neighbours (see this article in the Volkskrant of 23 February 2024).

The example of ‘The Friendly Doorbell’ coalition represents a first preliminary answer to our research question on how public safety can be co-created by citizens and authorities in an ethical and legal way. Other questions include whether the images of the public space – which such private doorbells often record, even though the law does not allow it – should be shared with the police or other authorities in case they record criminal activities, and where (and for how long) they should be stored. For example, should producers of such cameras be allowed to use the data to train their algorithms and thereby make more profit later on? And can producers adjust the design so that the doorbell only starts recording when someone actually presses on the bell?

Many questions are still outstanding, but we concluded that the possibilities of new technologies such as AI are only becoming clear once they are being implemented, and that they can change faster than any principle or law can capture. A big concern in particular is to whom and how the users of such technology should be accountable to. People within public institutions who are buying digital technology are not always aware of the technical features and data issues – but such awareness is necessary to bring out the ethical and legal issues which can then be discussed, and to inform councils and citizens properly.

We therefore end this blog post with a call to all involved in producing, buying, and using digital technology (and AI-equipped technology in particular) for public safety to hold structured conversations with each other about the objectives, implications, and ethics around it. These will help to raise awareness and safeguard the public values enshrined in the principles, with the aim to improve current and develop future applications and their regulation. The researchers involved in the AI-MAPS project look forward to continue their facilitation and contribution to such conversations together with project partners.

Related content
Blogpost for the AI-MAPS project by Ekaterina Voynova
ELSA magazine
Blogpost for the AI-MAPS project by Nanou van Iersel
Helix engagement with surveillance AI
Related links
Overview blogposts | AI Maps

Compare @count study programme

  • @title

    • Duration: @duration
Compare study programmes