The much anticipated European Parliament IMCO-LIBE Committees’ draft Report on the EU’s Artificial Intelligence (AI) Act was released on 20 April 2022. The report includes proposed amendments to the original text proposed by the European Commission.
Earlier this year, ECNL had published a set of proposals within our ongoing advocacy towards a EU regulatory framework for AI and human rights. One of our proposals called for amendments to ensure an accessible and effective mechanism for stakeholder engagement in context of the implementation of the AI Act.
We are pleased to see that the IMCO-LIBE Report acknowledges the importance of involving stakeholders and civil society organisations in the AI Act’s governance and implementation. The report largely reflects our overall advocacy to include opportunities for relevant external stakeholders and civil society to contribute to the tasks of the relevant EU and national authorities reviewing and implementing the AI Act (see our proposal).
We welcome, in particular, the new proposed amendments introducing:
- The possibility for users to “report to the relevant authorities or, where relevant, to the Commission, any serious incidents or breaches to national and Union law protecting fundamental rights resulting from the use of their AI system when they become aware of such serious incidents or breaches” (Amendment 39 to recital (78), taking stock of the concern we had expressed in our proposal of where relevant stakeholders can raise and discuss their legitimate concerns within the competent authorities to ensure they effectively exercise their tasks of supervision and implementation of the AI Act;
- The obligation for Member States to establish a central contact point for communication not only with AI systems’ operators but also for “other stakeholders” (Amendment 212 – Article 59, para 7) – as also proposed by ECNL;
- The “appropriate involvement and consultation of stakeholders and civil society“ in the regular evaluation and review of the list of high-risk AI systems (Amendment 18 – Recital 26b);
- The need for a “balanced representation of interests” in the development of standards and standardization by encouraging “the participation of all relevant stakeholders” (Amendment 33 - Recital 61, also reflected in Amendment 161 – Article 40, also reflected in Amendment 161 – Article 41, para 2);
- The need for the Commission to hold a public consultation (though with the ambiguous caveat “when relevant”) with “representatives of groups on which an AI system has an impact […] and civil society organisations” whenever it has to assess whether an AI system poses a “risk of adverse impact on fundamental rights” that is equivalent or greater than the one posed by the high-risk AI systems already listed in Annex III (Amendment 87 - Article 7, par 2a);
- The AI Board’s regular consultation (“twice a year”) and work with external stakeholders that include “relevant civil society organisations such as non-governmental organisations (NGOs), consumer associations, and the social partners and academia” on “issues related to the implementation and the effectiveness of this Regulation, regulatory gaps or loopholes observed in practice” (Amendment 37 - Recital 76 a, also reflected in Amendment 196 – Article 57, para 3c and Amendment 197 – Article 57, para 4);
- Ultimately, the need for the Commission to consult “relevant external stakeholders, in particular those potentially affected by the AI systems, organisations representing their interests, academia, the social partners and civil society” whenever conducting evaluations and reviews on the relevant AI Act provisions (Amendment 280 – Article 84, para 6).
What is next?
On 11 May, a joint IMCO-LIBE Committee hearing will consider the draft report on the Commission’s proposal for an AI Act. The competent shadows and opinion-giving committees will also be invited to contribute to the debate. The deadline for amendments to the draft report is set for 18 May 2022.
How to further improve stakeholders’ and CSOs’ participation in the AI Act
In line with the strengthening of stakeholders and civil society’s participation in the AI Act provisions on governance and implementation already tabled in the IMCO-LIBE draft report, we call on the Members of the EU Parliament to include an advisory group composed of relevant external stakeholders (including civil society organisations and representatives of affected groups) in the structure of the Board and its supervisory tasks.
We believe that setting up an advisory group would streamline multi-stakeholder engagement within the Board: for one, it would allow quicker feedback routes to the Board regarding the application and implementation of the AI Act. Furthermore, it could assist and advice on the way to reach out, e.g., to marginalised or non-organised groups impacted by AI systems in a meaningful way, thereby serving as a ‘bridge’ between the Board and broader civil society and other stakeholders.
In other words: the establishment of an advisory group of external stakeholders and CSOs to the AI Board will help operationalise all the well-received amendments in the IMCO-LIBE Report about meaningful stakeholder engagement and mitigate risks that consultations with relevant stakeholder are reduced to mere box-ticking exercises.