Draft EU rules would keep remote biometric identification under government thumb
The European Union appears ready to pick a side when it comes to using AI for mass surveillance and social credit rankings.
Draft legislation published by Bloomberg indicates that politicians will ban many AI uses, including most deployments of remote facial biometrics and other indiscriminate biometric surveillance roles. Violators would face a fine as large as four percent of global revenue.
The proposed rules, which could become law next week, would apply to any organization operating in the EU.
High-risk AI is spotlighted in the document. Special attention would be paid to software that could undermine the EU’s democracy and endanger the safety, lives and rights of its citizens.
Applications judged to be high-risk would be subject to pre-deployment inspection. Inspectors would make sure the algorithms were trained on unbiased data and with human oversight.
There would be exceptions, according to Bloomberg.
For example, a blanket exemption reportedly would be given to entities writing AI expressly for the military. And some remote biometric snooping would be allowed with special authorization.
The news publisher said proposed rules would, with some security exceptions, also make illegal AI created to “manipulate human behavior, exploit information about individuals or groups of individuals.”