EU Proposes Heavy Regulation of ‘High Risk’ AI Use

Organizations using AI in the EU could soon be required to be audited and subjected to restrictions on how it might be deployed. The use of AI for “social scoring” would be banned outright, while heavy restrictions will be imposed on a range of “high-risk” AI systems – though exceptions will be made for fighting serious crime.

High risk AI

The Artificial Intelligence Act proposal focuses on AI usage that might cause harm to EU rights and interests. Companies that do not comply could be fined up to €20 million or 4 percent of their turnover, reports Politico, which saw a draft of the proposed regulations.

As reported by CPO Magazine, “An AI application categorized as ‘high risk’ would be subject to special inspections, including examination of how its data sets are trained.” These could include financial applications, college admissions, and employment, among others.

Critics argue that the draft legislation does not adequately address discriminatory applications of AI technology such as predictive policing and migration controls at borders. Though organizations deploying them will need to give notice of their use and provide transparency on how the algorithms make their decisions, their use is still allowed.

Not strict enough

The European Data Protection Supervisor (EDPS), an independent body that monitors organizational processing of personal data expressed disappointment on the continued use of remote biometric identification systems.

“[The] EDPS regrets to see that our earlier calls for a moratorium on the use of remote biometric identification systems - including facial recognition - in publicly accessible spaces have not been addressed by the Commission,” it said in a press release.

“A stricter approach is necessary given that remote biometric identification, where AI may contribute to unprecedented developments, presents extremely high risks of deep and non-democratic intrusion into individuals’ private lives.”

The proposals are likely to be fiercely debated by both the European Parliament and member states, according to The Financial Times. Assuming it passes muster with both groups and is passed into law, the legislation is expected by 2023 at the earliest.

How the proposed rules will clash with the major EU goal of boosting its prowess in technologies such as AI remains to be seen, however. The commission has also published a road map to increase investments in AI and initiate various AI-centric strategies – such as pooling public data across the bloc to train machine learning algorithms, to help the EU fully benefit as a first-mover adapter.

Image credit: iStockphoto/PhonlamaiPhoto