The European Commission presented a new legislative proposal today including specific rules and actions to promote the development of artificial intelligence (AI). The regulation provides a first-ever legal framework as well as a 'Coordinated Plan on AI' for all the political changes and investments necessary for Member States in order to promote AI. This represents the continuation of the European AI strategy published in 2018 and aims to ensure the safety and fundamental rights of citizens and businesses while advancing AI as well as respective investments and innovations at a European level.
This proposal takes a risk-based approach categorizing AI systems as follows:
-
Unacceptable risk: AI systems or applications that manipulate human behavior in order to circumvent users' free will, e.g., toys with voice assistants encouraging minors to behave in a dangerous manner, or enable social scoring, e.g. authorities evaluating individual social behavior, will be banned.
-
High risk: AI systems that are used, inter alia, in critical infrastructures (transport), educational and vocational training (scoring of exams), private and public services (credit scoring) or in the administration of justice and democratic processes (application of the law to specific circumstances), should fulfill strict requirements – such as adequate risk assessment and mitigation systems, high quality of the data, detailed documentation and appropriate human supervision to minimize the risks.
-
Limited risk: Special transparency obligations apply when engaging with AI systems such as chatbots. Consumers should be informed that they are interacting with machine so they can consciously decide for or against doing so.
-
Minimal risk: This affects the vast majority of AI systems such as AI-supported video games or spam filters that pose minimal to no risk towards citizen's rights and their safety – their use should be free and without regulatory intervention.
Klaus Müller, President of the German Federation of German Consumer Organizations (VZBV), commented: "The VZBV welcomes the fact that the Commission is planning to introduce rules for AI applications. Unfortunately, the approach is feeble and despondent: it focuses only on a limited range of high-risk systems. All other systems, including those with a 'medium' risk, are neglected. This leaves consumers unprotected in many respects. (...) In terms of transparency, the proposal is also disappointing: merely labeling requirements are envisaged, which just apply to emotion recognition systems or when AI interacts with people. But in order for consumers to be able to exercise their rights in a self-determined manner, they must have a lot more information. This also includes information on risks, accuracy of errors and the data basis for a decision. Instead of an independent control body, users should largely check themselves whether high-risk AI systems comply with quality specifications and are legally compliant. That damages trust in and acceptance of AI."
Monique Goyens, Director General of the European Consumer Organization BEUC, commented: "Artificial intelligence technology already appears in the form of virtual assistants, self-driving car features and customer service chat bots. Today’s proposal misses the mark when it comes to protecting consumers. It is paramount this proposal gets improved, so consumer protection stays apace with technological advances. People should be able to trust any product or service powered by artificial intelligence, be it ‘high-risk’, ‘medium-risk’ or ‘low-risk’. The EU must do more to ensure consumers have enforceable rights, as well as access to redress and remedies in case something goes wrong."
Source: EC, BEUC & VZBV
More information, VZBV press release and BEUC press release