How Algorithms Threaten Our Freedom of Expression, and What the EU Can (and Should) Do About It

Supervisors: Prof. Dr. Bruno de Witte and Prof. Dr. Marta Pertegás Sender

Keywords: content moderation, algorithm, online platforms, digital services, freedom of expression

Online platforms increasingly use algorithms to identify and restrict prohibited content (i.e. illegal content as well as content violating the platforms’ own terms and conditions). When interviewed in 2018, Mike Schroepfer, former chief technology officer at Meta Platforms, even made a striking confession‘To me, AI is the best tool to implement the policy – I actually don’t know what the alternative is’. This narrative is also perpetuated by some EU institutions seeking to encourage online platforms to enhance their fight against illegal third-party activity. However, scholars, civil society actors, and users of online platforms repeatedly point out that algorithms are opaque, have low accuracy, and can reproduce harmful biases. In light of these pressing concerns, could a proactive, algorithm-driven approach to content moderation ever be reconciled with due respect for freedom of expression?

My doctoral thesis ‘Silenced by Default: Algorithmic Content Moderation and Freedom of Expression in the European Union’ has examined how the advent of algorithms is transforming EU platform regulation and presenting unique challenges for users relying on online platforms to publish or consume content. It argues that if not subject to appropriate safegaurds, algorihtmic content moderation can severely undermine the quality and diversity of public debate online. The EU should therefore avoid endorsing the use of algorithms where it can put freedom of expression at risk. Additionally, it should stimulate a responsible deployment and development of algorithms in content moderation even when implemented by online platforms on a voluntary basis. The adoption of the Digital Services Act in 2022 (which some call ‘the European constitution for the Internet’) is certainly a big breakthrough for the protection of freedom of expression in the online environment. Yet more work is needed to enhance transparency, contestability, and trustworthiness of algorithmic content moderation. The thesis provides a policy roadmap which would hopefully inspire a thoughtful policy response to this complex issue.

The thesis also reflects on broader dilemmas of platform regulation in the EU. For instance, online platforms are often considered ‘gatekeepers’ of online communications, which justifies a more stringent regulatory treatment. However, the thesis shows how the recent reforms of the EU legal framework governing online platforms can inadvertently enhance rather than constrain their economic power and societal influence. The thesis also addresses the limitations of EU fundamental rights law in countering the threats to freedom of expression posed by the actions of private tech companies. Therefore, the thesis makes a contribution to multidisciplinary scholarship on law and technology.