How the European Parliament is shaping future AI legislation
The European Parliament's special Committee on Artificial Intelligence in a Digital Age (AIDA) concluded its work this month (April 2022) with the adoption of its final report, investigating the respective challenges that will come with the deployment of AI technologies and concluding with an urgent call for EU action.
The uptake of Artificial Intelligence (AI) systems has been recognised as a key priority by the EU, which, in 2018, called for an "Artificial Intelligence for Europe". Since then, EU policy makers have been trying to define what would constitute a European approach to AI and establish Europe's position as a global hub for excellence in AI.
It was commonly acknowledged that while AI can significantly improve people's lives and bring major benefits to society and the economy, it also raises specific concerns with regards to safety, security and fundamental rights protection. Therefore, in April 2021, the EU took the first steps towards regulating Artificial Intelligence technologies and managing the associated risks. Experts, technology leaders, civil society and other stakeholders have been bringing forward a number of concerns and calling for amendments, seeking in particular to reconcile democratic oversight of the design and implementation of EU AI regulation with the need to encourage innovation.
In this article, we take a closer look at the European Parliament's:
- non-legislative initiatives
- work on the Commission's proposal for an EU regulatory framework on AI - key dates and key actors
- In October 2020, ahead of the European Commission proposal, the European Parliament has adopted three reports which include Recommendations on an ethics framework for AI, a civil liability regime for AI and intellectual property rights for the development of AI technologies.
- In January 2021, the European Parliament proposed guidelines for military and non-military use of AI, especially in areas such as military, justice and health.
- In May 2021, the European Parliament adopted a resolution on the use of AI in education, culture and the audiovisual sector. The resolution underlines the need to train AI technologies using unbiased data sets with the aim to protect non-discrimination, gender equality, pluralism, as well as cultural and linguistic diversity. The Parliament has also adopted a Resolution on Europe's digital future and the use of AI applications for consumers.
- Finally, in October 2021, the European Parliament with a Resolution has opposed the use of AI technologies by the police and judicial authorities in criminal matters and referred to the risks in relation to the protection of people's fundamental rights.
In June 2020, the European Parliament decided to set up a special Committee on Artificial Intelligence in a Digital Age (AIDA) to reflect on the future impact of AI technologies. It examines six case studies:
1) AI and health
2) AI and the Green Deal
3) External policy and the security dimension of AI
4) AI and competitiveness
5) AI and the labour market
6) AI and the future of democracy.
The Committee concluded its work in April 2022 and adopted its final report which investigates the respective challenges that will come with the deployment of AI technologies and concludes with an urgent call for EU action.
The European Parliament's work on the Commission's proposal for an EU regulatory framework on AI
- AI Regulatory sandboxes (Article 53, AM 174): The co-Rapporteurs propose greater flexibility for companies with the way sandboxes are being designed to offer room for innovation and experimentation. "The AI regulatory sandbox shall allow and facilitate the involvement of notified bodies, standardisation bodies, and other relevant stakeholders when relevant." This appears to be in line with industry's preferences which are against an one size fits all approach.
- Free of errors data sets for high-risk AI systems (Article 10, paragraph 3, AM 96 and recital 44, AM 29): The co-Rapporteurs recognise that the error free requirement will be challenging for industry to implement and an excessive requirement therefore they have proposed a different formulation i.e. "to the best extent possible, taking into account the state of the art, free of errors" datasets.
- Forbidden practices (Recital 17a, AM16, Article 5, paragraph 1, AM 76): The co-Rapporteurs have agreed to add practices that amount to "predictive policing" to the list with the view to protect the key principle of presumption of innocence established in our liberal societies. This seems to be a difficult point for agreement as the European Commission sees some benefits in helping victims or people to protect themselves, while the Rapporteurs and a number of NGOs like European Digital Rights call for a total ban. The Council has proposed an amendment to extend the prohibition of using artificial intelligence for social scoring also to private actors, not just for the public sector. This would result in prohibiting the use of AI in making classifications of citizens for illegal uses, something which is occurring in other parts of the world.
- List of high-risk applications expanded (Annex III, paragraph 1, AMs 287-297): AI systems used to influence or shape children development such as in educational and vocational training; AI systems for elections and votes counting; AI systems for the triage of patients in the healthcare sector, and AI systems used to determine eligibility for health and life insurance.
- New provisions - right to redress (Recital 84 b (new), AM 46 and AMs 268-270, 280): Individuals can lodge a complaint against AI providers and receive compensation for damage suffered. Non-compliant providers can be fined.
- Stronger role of stakeholders: The co-Rapporteurs have proposed an increased involvement of stakeholders and civil society organizations with regards to different provisions for example the updates to the list of high-risk AI systems, the standardization process, the activities of the Board and the sandboxes. The AI Board should organize, at least twice a year, consultations with industry, start-ups and SMEs, civil society, and academia, in order to carry out its tasks in collaboration with all relevant stakeholders
- Deadline for Amendments to the draft report is June 1 at noon
- Discussion at Telecom Council June 3
Provisional timeline for the Artificial Intelligence Act
Provisional timeline for the Artificial Intelligence Act
How can we help?
To find out more about shaping future AI legislation, get in touch with our Public Affairs team.