Lawmakers in the European Parliament today (13 March) approved the AI Act, rules aimed at regulating AI systems according to a risk-based approach.
Category
🗞
NewsTranscript
00:00 "The European Parliament has approved the first artificial intelligence law in the world
00:06 by overwhelming majority. The text sets a risk-based approach that goes from marking
00:12 when images have been generated by AI to completely banning some uses. Some of those are systems
00:19 that use ethnical, religious or facial traits to identify people or the recognition of emotions."
00:26 "We prohibit the emotional recognition of workers and students. There will be no possibility
00:32 for an employer to look for the emotions of the employee, which is something that we think
00:39 is not acceptable in Europe."
00:43 But some of those bans will have exceptions if used by security forces. This was the biggest
00:49 disagreement between Parliament, who aimed for a complete prohibition, and the member
00:54 states, wanting an exemption to use those systems in the case of a terrorist attack
00:59 or human trafficking.
01:00 "We've introduced judicial redress and judicial authorization as a mandatory step in allowing
01:08 for this use of technology. We've limited the scope of use to a very strict set of crimes,
01:17 the most serious crimes, to make sure that again law enforcement would not be tempted
01:22 to deploy this technology in public spaces all the time."
01:26 But some MEPs think the law legitimizes the misuse of AI technology.
01:31 "No terrorist has been found using real-time mass surveillance, but it has a very chilling
01:38 effect. It discriminates against minorities because these are often times wrongly reported."
01:47 After a formal approval by member states, the law will be published around May. In November,
01:55 bans will enter into force and by 2027, the AI Act will be fully functioning.
02:00 (whooshing)