AI technologies are slowly becoming ubiquitous, with applications in fields as diverse as law, manufacturing and education. These developments go hand in hand with changes in infrastructure (e.g. for data) and in organisational processes (such as platformisation), while also shifting ideas of privacy, responsibility, accountability, property or even labour. The context of AI is exceptionally broad, and for our female experts at Women in AI Austria, 2021 offered many opportunities to comment on regulatory proposals relating to this digital technology. A special thanks at this point to the dedicated policy team of Alexandra Ciarnau and Rania Wazir, without whom this work would not have been possible!
Digital Services Act (DSA) and Digital Markets Act (DMA)
The Digital Markets Act and the Digital Services Act proposed by the European Commission last year will have a significant impact on how AI technologies can be used on platforms (DSA) and how companies providing AI services can access the European market, whether as providers of services using AI or as business users of regulated services – especially cloud services (DMA). We see both legislative instruments as a vital package and welcome these efforts to increase fairness and accountability, both for private users and businesses. We particularly support the access to data for researchers foreseen in the DSA, although we have some suggestions on how to improve it to account for civil society research. We also offered some suggestions on how the DSA could provide for the development of third-party recommendation systems for users of (social media) platforms, which could be an important step for countering the harms caused by proprietary recommendation systems and offer more choice for users, especially regarding the type of content they would like to give priority to. In our view, third-party recommendation systems could be further strengthened by provisions in the DMA. One major weakness of the DMA in our view is the lack of a complaint procedure, which end users (and the organisations representing them) as well as businesses could turn to in case of persistent problems when dealing with gatekeepers. Check out our responses to learn more about our thoughts on the DSA and the DMA.
Data protection and AI toolkit
Despite all the talk about ethical AI, concrete tools helping companies to assess their systems and evaluate whether they comply with legal (and/or ethical) requirements are comparatively rare. When we heard that the Information Commissioner’s Office (UK) designed a toolkit for assessing and mitigating risks to data protection created or exacerbated by AI which processes personal information, we decided to take a look and see whether this could be a useful tool for our members. For those of you concerned whether you might need to introduce additional measures to ensure that the AI system you are developing does not cause harm using personal data, we think that this is an excellent tool to get started on this topic!
Transparency of political advertising
Political advertising is somewhat of a double-edged sword for democratic elections: on the one hand, advertising campaigns can help voters become aware of the entire range of candidates, but on the other hand, targeted online advertising campaigns exploiting people’s interests can be a serious threat to a voter’s ability to make an informed choice. Rania Wazir, one of our policy team members and member of the board of the Vienna Data Science Group, is one of the few experts on this topic in Austria, having previously cooperated with Wahlbeobachtung.org to monitor political advertising during the Austrian elections. Drawing on her experience and expertise, we submitted a response to the European Commission’s proposal for an initiative to ensure greater transparency of sponsored political content.
The Council of Europe’s Ad Hoc Committee on AI launched a multi-stakeholder consultation in April 2021, to which we had the pleasure of responding. Amongst other things, we called for basing the regulation of AI on the legal principles of respect for human dignity, equality, freedom of expression, personal integrity and the possibility to challenge a decision made by an AI system (and access to an effective remedy) – the choice was, however, a tough one. One of the important legal gaps which could be addressed at the level of the Council of Europe was the prohibition of mass surveillance as well as a ban of military AI applications – which, to our chagrin, was not foreseen in the AI Act either.
UN Human Rights Council panel discussions
In June, the Human Rights Council hosted two panel discussions centring on its role in promoting a human rights-based approach to new and emerging technologies. Our President, Carina Zehetmaier, happily accepted the invitation to attend and made a case for the development of human rights impact assessments to reduce bias in AI systems as well as biased and unjust outcomes for humans.
People vs. Big Tech
Remember our position on the DSA? Well, we share some of our views with other European organisations and signed the People’s Declaration together with 112 other European organisations. Human rights are important cornerstones of our society, and the digital transition will only be sustainable if we ensure that human rights are cornerstones of our digital society.
After ending 2020 with an article explaining our position on the consultation of the #SAIFE paper of the OSCE, we decided to use the AI Act as another opportunity to get creative. With a complex piece of legislation such as the AI Act, it is important that the interactions between the articles, the extent and responsibilities for the roles and the processes required are easy to understand for anyone wishing to participate in the discussion. Our posters on the AI Act try to do exactly that: help people find their way around the legislative proposal to have a voice in the debate. We also created a briefing for those who prefer the long form, and of course we also submitted feedback on the proposal itself – head over to our blog post on the AI Act to learn more!
Declaration of Digital Principles
The Declaration of Digital Principles was recently published and will hopefully be signed by the EU Member States in the course of 2022. Before the document was finalised, however, we decided to submit our two cents on the proposal in September 2021. Later that month, we were invited to attend a stakeholder workshop where we argued that AI can help ensure equal access to public services throughout the EU and that terms and conditions, especially relating to AI services, are crucial for ensuring transparency and fair treatment of consumers and businesses in the EU.
Austrian AI strategy (AIM AT 2030)
Last but certainly not least, the long-awaited Austrian AI strategy was published in late August, setting out overarching aims for the developing AI landscape in Austria and specifying sectors or areas of application where AI development could be particularly useful. While there was no opportunity to comment on the document, this strategy is important for our own strategic goals: namely, helping women find their way into the field of AI in Austria, providing a platform for them to showcase their expertise and helping to achieve a societally sustainable and inclusive field.
Valerie Hafez is a member of the board of Women in AI Austria and leads our policy initiatives. An anthropologist by training, she is currently studying science and technology studies at the University of Vienna.