Panoptic Brand Protection? Algorithmic Ascendancy in Online Marketplaces

Abstract

As e-commerce continues to grow, so do counterfeits, which circulate online in ever greater numbers. Larger online marketplaces - notably, Amazon, eBay and the Alibaba group - have responded by developing preventive, proactive filtering tools powered by machine learning algorithms. These automated filters detect and remove counterfeits at scale. While the legitimacy of proactive algorithmic filtering in the copyright context – and Article 17 of the Digital Single Market Directive in particular – has been extensively debated in the EU, the voluntary adoption of such filtering technology for trade mark protection has gone largely unremarked. Drawing on relevant EU law, this article seeks to fill this gap by (1) outlining the prevalence of counterfeits in e-commerce; (2) detailing the extent to which ML-enabled filtering technology has been developed in response; (3) identifying the drivers for its voluntary adoption, despite filters not being a necessary requirement to shield marketplaces against secondary liability claims; before (4) analysing whether the recently enacted EU Digital Services Act (DSA) offers process-based remedies, for the ills of a system of “privatised injunctions” enabled by algorithmic filters.

Keywords: trade mark, brand, machine learning, algorithm, enforcement, AI