EC Floats Rules for 'Trustworthy' AI, to Some Skepticism

The European Commission's "first-ever" draft rules for AI unveiled Wednesday prompted a mix of praise and criticism. "Trust is a must" for AI, and the rules will spearhead "the development of new global norms to make sure AI can be trusted," said Margrethe Vestager, EC vice president for a Europe fit for the digital age. The rules take a risk-based approach. AI systems deemed a clear threat to users' safety, livelihoods and rights would be banned, including AI that manipulates human behavior to circumvent users' free will and systems allowing government "social scoring." High-risk uses include those involved in critical infrastructure; those that could put people's lives at risk; safety components; and law enforcement. High-risk systems would be subject to strict obligations, including adequate risk assessment and mitigation; logging activities to ensure results are traceable; and giving users clear, adequate information. "All remote biometric identification systems are considered high risk and subject to strict requirements." Limited-risk AI systems such as chatbots would have specific transparency conditions, like notifying users they're interacting with a machine. Low-risk uses such as AI-enabled videogames or spam filters, which comprise the majority of AI systems, won't face the regulation. The EC proposed that national market surveillance authorities supervise the rules and that a European Artificial Intelligence Board be established. The plan foresees voluntary codes of conduct for non-high-risk AI. It needs the European Parliament's and EU governments' OKs. Parliament's Special Committee on Artificial Intelligence in a Digital Age welcomed the proposal. Lawmakers now "need to act on two fronts," said Chair Dragos Tudorache of Renew Europe and Romania: Reduce unnecessary burdens on startups, small and midsize businesses, and industry so "AI can be unleashed to its full economic potential" and boost citizens' rights. The Computer & Communications Industry Association applauded the risk-based approach, saying the proposal should be "clarified and targeted to avoid unnecessary red tape for developers and users. ... Regulation alone will not make the EU a leader in AI.” It's a "bold step towards pioneering regulation in this field," said the Information Technology Industry Council, urging the EC to focus on flexible rules targeted to the highest-risk applications. BSA|The Software Alliance urged the EC to engage with international partners, since building trust in AI is a shared responsibility. Others were less enthusiastic. The draft fails to prohibit "the full extent of unacceptable uses of AI," particularly biometric mass surveillance, and allows too much industry self-regulation, said European Digital Rights. The Center for Data Innovation accused the EC of striking "a damaging blow to the Commission’s goal of turning the EU into a global AI leader by creating a thicket of new rules." The recommendations overly focus on too limited a range of AI uses, said the European Consumer Organisation: It omits many uses that affect people's everyday lives, such as smart thermostats, and doesn't ensure consumers have enforceable rights.