Europe's Regulatory Power Play May 2021
Subscribe to Insights in Brief to be notified about new Featured Content as it becomes available!
The European Commission has published proposals for a regulatory framework for AI. The proposed AI regulation follows a risk-based approach (as does, for example, regulation of medical devices) to create requirements for high-risk AI systems that are different from requirements for lower-risk AI systems. High-risk AI systems include those that serve critical infrastructure, product-safety components, and myriad applications in which AI risks discriminating against people (including in recruitment, credit-scoring, law‑enforcement, and judicial applications). In high-risk cases, the regulators propose that developers should train AI with "high-quality datasets" and log activity to "ensure traceability of results," among other measures. Lower-risk AI systems that interact with humans should ensure transparency (for example, a chat bot should be clear that it is not a person). The proposals suggest free use of minimal-risk AI systems (such as within video games). Penalties for violations of the regulations are significant: €30 million ($36 million) or 6% of global revenues—whichever is higher.
The proposals join other new ambitious plans for digital regulation in Europe. The forthcoming Digital Services Act (DSA) and the accompanying Digital Markets Act (DMA) represent the first major overhaul of the European Union's rules for governing the internet in two decades. The DSA creates new responsibilities for online intermediary services (including Big Tech platforms), and the DMA aims to prevent "gatekeeper" platforms from imposing conditions that the European Union judges unfair to businesses or consumers.
The European Union's General Data Protection Regulation set the tone for more stringent data regulation around the world. European regulators are now making a variety of moves to become the world's de facto regulators of AI, big data, and digital markets in general (for example, vendors are unlikely to want algorithms for the United States that are different from algorithms for Europe). The moves will affect Big Tech firms such as Google and Facebook that benefit from large data holdings and the rollout of data-driven AI. The AI regulations, the DSA, and the DMA still have more than a year of approval cycles to work through. Tweaks are possible, but major reworkings are unlikely, despite protests from some US Big Tech firms.
Globally, regulation of digital markets is increasing, but Europe is in a position different from that of the United States and China because it has no Big Tech firms to protect. For the United States and China, regulating digital markets is a balancing act. The government of China recently fined Big Tech player Alibaba $2.8 billion for anticompetitive activities. Although this fine is significant, it is actually much lower than the maximum fine that Chinese law allows—high enough to flex government muscle but not high enough do real damage. Similarly, US regulators see increasing needs to regulate Big Tech but also know that US economic growth is dependent on Big Tech progress. In contrast, Europe has incentives to use regulation to protect its markets.