Research exempt from proposed EU ban on ‘unacceptable’ AI

Ethical studies of artificial intelligence systems that “distort” human behaviour will be allowed

The European Commission has proposed EU laws that categorise different uses of artificial intelligence according to the risk involved, with an outright ban on uses entailing an “unacceptable risk” but an exemption from the ban for some forms of research.

Under the proposals unveiled on 21 April, intended to make Europe “the global hub for trustworthy AI”, unacceptable uses of AI would be any considered to pose “a clear threat to the safety, livelihoods and rights of people”, such as systems that “manipulate human behaviour to circumvent users’ free will” or that “allow ‘social scoring’ by governments”.