Commission presents proposal for EU legislation on Artificial Intelligence
The European Commission has presented its proposal for EU legislation on artificial intelligence under the legal bases of Article 114 and Article 16 TFEU. The new rules follow a ‘risk-based’ and ‘human-centric’ approach and offer a ‘future-proof definition of AI’.
Scope of the proposed rules
The new rules would cover both public and private actors inside and outside the EU as long as the AI system is placed on the EU market or its use affects people located in the EU. It can concern both providers and users of high-risk AI systems (defined in the proposal), who will have to carry out conformity assessments or install quality and risk management systems. It would not apply to private, non-professional uses.
Division of AI into risk categories
That reference to ‘high-risk AI systems’ comes from a division of AI in the proposal into four risk categories to be in line with EU fundamental rights law, which includes a definition and methodology for determining that risk.
The definition for ‘artificial intelligence system’ is ‘software that is developed with one or more of the techniques and approaches listed in Annex I and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with’.
There is first, ‘unacceptable risk’ (Article 5) –
- an outright ban on AI that is considered a clear threat to the safety, livelihoods and rights of people;
- an outright ban on AI that allows ‘social scoring’ by governments.
Second, it distinguishes areas in which AI can by used by the level of risk posed by AI, with that in the high-level risk category (Title III, Chapter 1) being subject to extra conditions, for example:
- Critical infrastructures such as transport;
- Educational or vocational training that may determine the access to education and professional course of someone’s life;
- Safety components of products (such as in robot-assisted surgery);
- Employment, workers management and access to self-employment;
- Essential private and public services (such as loan applications);
- Law enforcement that may interfere with people’s fundamental rights;
- Migration, asylum and border control management;
- Remote biometric identification in general; and
- Administration of justice and democratic processes.
And the other risk levels are ‘limited’ and ‘minimal’ risk. The Commission’s press release describes most AI systems as falling in the minimal risk category, and are therefore not covered by the proposed new rules (though voluntary codes of conduct (Title IX) and regulatory sandboxes are suggested).
National competent authorities as the enforcers, and a new Board to facilitate implementation
In order to oversee the governance of AI regulation, the proposal suggests this be carried out by the supervision of national competent market surveillance authorities (NCAs) – it will be for Member States to identify which national authority is the best placed (Title VI, Chapter 2).
It also suggests the creation of a European Artificial Intelligence Board (Title VI, Chapter 1) made up of those NCA representatives, the European Data Protection Supervisor, and the Commission to facilitate implementation of the rules. The Board would, inter alia, issue recommendations and opinions to the Commission regarding high-risk AI systems and on other aspects relevant for the effective and uniform implementation of the new rules.
The new rules propose sanctions (Article 63) in the case of persistent non-compliance, such as a fine of between 10-30,000,000 euros (depending on the infringement) or up to 6% of the annual turnover of an offending company.
EU co-legislators now to consider the legislative proposal
The European Parliament and the Member States must now decide whether to adopt the Commission’s proposals: the Commission has in parallel proposed a revised coordinated plan with Member States (that takes into account the EU’s digital and green transitions); and a new Machinery Regulation to ensure that the new generation of machinery products guarantee the safety of users and consumers, and encourage innovation) in the ordinary legislative procedure.
EU Law Live will be publishing a series of Insights on EU law and Artificial Intelligence, and on the ERA’s excellent three-day conference on ‘Human Rights and Artificial Intelligence Systems’, held on 29-31 March 2020.