The EU is poised to ban AI for mass surveillance or for ranking social behavior.

A draft set of proposals obtained by Bloomberg shows that the EU could fine AI and tech companies as much as 4% of global revenue if they fall foul of the EU’s new norms for the use of AI. Further, these proposals include a ban on AI-powered mass surveillance or systems that rank social behavior.

The European Commission could launch the new measures as soon as next week.

The EU’s position on AI would therefore be radically different from the US and China. The US prefers to leave the giant tech companies to self-regulation; the other extreme is China, which is reportedly harnessing AI in its drive to become a surveillance state.

Features of the new regulations

The draft document proposes the following:

• a ban on the use of AI for indiscriminate surveillance
• a ban on AI systems that determine social credit “scores” and attempt to evaluate somebody’s trustworthiness based on their behavior or personality traits
• notifications to people whenever they interact with an AI system
• the use of remote biometric identification systems (such as facial recognition in public) only with prior special authorization
• special monitoring of high-risk AI systems such as self-driving cars (personal safety); systems for job hiring, judicial decisions, and credit scoring (livelihood)
• such high-risk AI systems to be assessed beforehand to ensure (1) they are explainable to human beings and (2) that the data sets used for their training are free from bias and of high quality

Also, the rules propose that EU member states would need to install assessment bodies having the responsibility to test, certify, and inspect AI systems.

AI companies that provide prohibited services, do not cooperate with the authorities, or supply incorrect or falsified information, could be fined up to 4% of their global revenue.

These proposals could change before their official announcement on April 21.

They do not apply to AI systems used entirely in military applications.

Related Story:  China’s Huawei And Megvii Tested AI To Pinpoint Uighur Muslims