Abstract

This talk circles around the complexity of automated content moderation on social media platforms, and to what extent this growing web of regulation allow or demand AI moderation, in relation to human content moderation.

The increased societal importance of social media with an enormous amount of content being created and shared between users – and consequently – the increase in the power of platforms, has in recent years been met with attempts of tackling the problem through European regulations. This regulation, the Digital Services Act being the focal point in Europe, is aiming at creating safer spaces online and a more transparent experience for users of social media. In turn, this has increased the responsibilities of platforms with demands to remove harmful or illegal content rapidly. As a result, this has created a situation where social media platforms, especially very large platforms, will need to use automated moderation technologies to successfully moderate an almost unimaginable amount of potentially harmful content in a limited timeframe. Not only can automated systems handle considerably more content, and do it faster than human moderators could, it can also ease the burden that human moderators face when being exposed to disturbing content. However, humans can make complex contextualized decisions and understand aspects like underlying intent or cultural circumstances, preventing undue restrictions in content and infringement of users’ freedom of expression.

Speaker

Therese Enarsson is a docent in law at the Department of Law at Umeå University. In her research she has over the recent years focused on fundamental rights in different legal contexts, such as the protection of victims online, and content moderation online, with a focus on protecting and balancing free speech and other rights. Her research spans over many legal fields such as criminal law and constitutional law, and she has also published interdisciplinary research in the social media context.

Please note. The subtitle was primarily generated by AI in collaboration with human review and may contain some errors.