U.S. states are exploring AI regulation inspired by the EU AI Act, focusing on limiting algorithmic discrimination. Colorado has initiated legislative efforts, with proposals emerging in multiple states. However, compliance burdens raised by these new measures could pose challenges for businesses, particularly in biometrics, where existing state proposals lack restrictions on real-time remote identification.
The U.S. is currently navigating the regulation of artificial intelligence (AI), showing a divergence in approaches between federal and state governments. Some commentators, including the Hyperdimensional Substack, suggest that state policies may be echoing the European Union’s AI Act. However, key differences persist, particularly regarding the regulation of biometrics, demonstrating a uniquely American stance on compliance and economic growth.
Colorado was the pioneer, enacting SB 205, though its implementation is lacking clarity. Other states including California, Connecticut, Iowa, and New York are also considering similar legislative measures aimed at reducing AI-related harms like algorithmic discrimination. These initiatives align with a common goal of conducting impact assessments to identify potential risks.
Dean Ball from George Mason University expresses concerns over the influence of the EU on U.S. regulations, coining the phrase “The Brussels Effect.” He observes that the algorithmic discrimination bills in various states mimic key elements of the EU AI Act, potentially leading to high compliance costs and increased litigation for U.S. businesses.
In contrast to the EU’s prohibition on real-time remote biometric identification in public areas, U.S. state proposals have not introduced such restrictions. The focus is rather on shifting accountability to ensure laws are adhered to before deploying technologies, which Ball describes as a “preemptive approach.”
Facial recognition developers are taking measures to address dataset imbalances earlier in their processes. However, comparative statistics on wrongful arrests associated with facial recognition are sparse, challenging the assumption that these technologies pose significant risks. Regulatory concerns at the state level primarily revolve around biometrics and AI, though opinions vary on the severity of these issues.
The regulatory landscape for AI and biometrics in the U.S. is evolving, with states adopting measures similar to the EU AI Act but diverging on key facets such as real-time biometric surveillance. While attempts to mitigate algorithmic discrimination are prominent, the actual implications for biometric technologies remain limited. The balance between compliance costs and economic opportunities continues to be a focal point in this legislative dialogue.
Original Source: www.biometricupdate.com
