In February 2020, the EU Commission presented the White Paper on Artificial Intelligence (AI), which forms the basis for the 2021 Commission proposal. Beforehand, the European Parliament published three legislative resolutions on the subjects of ethnicity, civil liability and intellectual property. The draft is now the topic of an intensive debate by the European Parliament and the Council. Recently, Parliament adopted a Resolution, which demands a ban on facial recognition software by law enforcement authorities, with a large majority. In June 2021, a Progress report was published under the Portuguese Presidency. Next week, 14 October 2021, the Regulation will be debated at ministerial level by the Telecommunications Council, whereby the debate will primarily focus on the regulatory design and the challenges of an effective implementation of the AI Act.
 

AI Regulation requires “human-centred approach“

AK EUROPA has analysed the AI Draft regulation of the Commission in a new Position Paper. Overall, it is to be welcomed that this future-oriented issue has been taken up to create a framework to promote, develop and use as well as to regulate AI. The degree of regulatory requirements shall be determined by evaluating the technology’s risk. However, the draft primarily focuses on the technology with a view to the requirements of the Single Market. The originally promoted “human-centred approach” hardly gets a mention. Many appropriate approaches are watered down by exceptions and restrictions. Moreover, the institutional setting of the authorities, charged with the Regulation, is open.

Artificial Intelligence and world of work insufficiently addressed

From the employees’ perspective, several aspects of the draft regulation are inadequate. The subject of “Artificial Intelligence and world of work” has not been addressed. Hence, necessary safeguards for affected employees with regard to using AI at the workplace are missing. Hereby, the following points are of special importance:
 

• Enshrining the (inter)corporate labour interest representation when introducing and using AI at the workplace as well as appropriate co-determination and veto rights of employees and their representations of interest.
• Applications, which may have a negative impact on work realities and conditions, have to be classified as “high-risk”, which would subject them to stricter regulations.
• Certain applications in relation to employment (automated decisions in individual cases and profiling) should be banned.
• A “bottom-up approach” is required, which from the start reviews the impact of AI on the workforce and its working conditions and which makes the final decision of its use only on the basis of these experiences.
   

Need for improvement from the point of view of consumers

From the point of view of consumers, the draft regulation also shows fundamental flaws. Consumers have to be protected as best as possible against their fundamental rights and freedoms being undermined, against non-transparency, discrimination, physical and mental risks and other risks of damage, which originate from AI und algorithms. The following points are required among others:
 

• Regulations must not only guarantee transparency, freedom from discrimination and rights of appeal in respect of high-risk AI, but also in case of “merely” risky applications.
• Enshrining rights for affected citizens and consumers, among others the right to information, self-determination and rights of appeal.
• Banning socially undesirable AI systems instead of fragmentary bans on some varieties of Social Scoring, remote biometric monitoring and behavioural manipulation.
• AI certification granted without exception by independent authorities instead of purely self-certification by manufacturers.
• Protection standards for biometric AI analysis in case of consumer business.
• Revision of outdated rules for product liability and product safety.
   

Even though it is to be welcomed in principle that AI shall be given a European legal framework, many necessary framework conditions for employees and consumers are missing. The “human-centred approach” falls by the wayside for the benefit of a liberal market for AI. It would be important for Europe to assume a leading role and to take the interests of citizens with regard to using AI into account.

Further information:

AK EUROPA Position Paper: EU-Regulation on Artificial Intelligence
AK EUROPA Policy Brief: The body as an access key? Biometric methods for consumers
AK EUROPA: Artificial Intelligence must guarantee consumer protection