LONDON - The European Commission’s recent draft artificial intelligence (AI) regulation ticks off obligations providers, manufacturers, distributors and users must bear, and spans the gamut of players in the AI chain.

Most of the  duties the rule issued on April 21 impose target high-risk AI systems, i.e., those forming part of products, or themselves products, subject to European Union law and include medical equipment, toys, elevators, pressure equipment, and those deemed high risk as a matter of law, such as for biometric identification and that of natural persons, education and vocational training, and employment. Management and self-employment, access to essential private services and public services and benefits, while law enforcement, migration, asylum and border control management also fall under their purview, as do administration of justice and democratic processes. 

The EU’s aim is to be seen flourishing an exhaustive list to project an aura of legal certainty.

The devoirs heaped on high-risk AI systems are indeed weighty, and are hedged with requirements, addressing risk, data governance, technical documentation and record-keeping, transparency and provision of information, human oversight, accuracy, robustness, and cybersecurity.
The EU AI Act also mandates set procedures for those groups tasked with conformity assessments, and sets forth various challenges to their decisions.
The Act is also heavy on transparency. The meaning of transparency, especially in relation to fairness and explainability, and how it must be seen at the junction of consumer and privacy law.

One area drawing the gaze of privacy vigilantes worldwide was that for systems scanning human traits, such as AI for emotion recognition or biometric categorization, those with something to hide felt that the Act was more about legitimizing intrusive AI rather than banning it.

On, June 22, the day a