Supporting the responsible use of AI in European local authorities: contract clauses for the procurement of AI
Invitation for a roundtable exercise to peer review common AI procurement clauses for local authorities.
Local authorities are increasingly using AI technologies and algorithmic systems to operate their territories and provide services to citizens. With AI having a bigger impact on citizens’ lives, there is a growing need for instruments that can guarantee the use of AI in a responsible and secure manner, ensuring for example transparency about the decision-making processes and accountability.
The European proposal for an Artificial Intelligence Act establishes a regulatory approach to AI that aims at ensuring safe, transparent, and trustworthy AI systems. In particular, the AI Act sets harmonised rules on AI systems using a risk-based approach that introduces mandatory requirements, conformity assessments and transparency obligations for the users and providers of AI systems, including public administrations.
In line with proposed regulation and to support its responsible use of AI, the city of Amsterdam developed in 2021 a set of contractual terms for algorithms that they purchase from suppliers. To have insight into the process observed during the development and application of the algorithm, the terms describe which information the city requires from the suppliers. Such contractual terms for the procurement of AI help to operationalise standards, create obligations, and establish responsibilities for trustworthy, transparent, and accountable development and procurement of AI technologies.
Building on the work done by Amsterdam, legal expert Jeroen Naves in collaboration with DG GROW and DG CNECT, has now updated the terms by Amsterdam and developed them into common AI procurement clauses. These clauses are in line with the latest version of the AI Act proposal and consider recent developments related to AI.
The ambition of this roundtable exercise is to peer review these clauses with a European audience and draft a standard for public procurement of AI and algorithmic systems that can be used by public authorities across the European Union.
Peer review approach
Objective: The objective of the roundtables to validate and improve the AI procurement clauses with the support of experts from public sector bodies, academia and supply side
Format: The clauses will be covered using two roundtables. Participants are expected to take part in both roundtables:
- Round Table 1: Thursday, 25 May 2023, 10:00 – 12:00.
- Round Table 2: Wednesday, 7 June 2023, 10:00 – 12:00.
The roundtables will be held online, through Microsoft Teams, and include discussions and break-out exercises using pre-defined formats and collaborative tooling.
Participation: Experts from legal, industry, technical, procurement and ethics domains are invited to participate by expressing their interest through the online form here. Registration is closed at Friday 19th of May.
Selection: Based on the the number of expressions and the professional background of the experts a selection will be made to make sure the feedback represents the various perspectives.
Preparation: Participants will receive a confirmation via e-mail about their involvement in the roundtables, including the updated version of the AI procurement clauses. To support an effective peer review exercise, everyone is expected to familiarise themselves with the procurement clauses before the roundtables.
Validation: The procurement clauses will be updated using the input that is collected during the two roundtables. The updated version will be circulated with all the participants for written validation.
Publication: After validation, the AI clauses will be translated in all EU languages and these will be made publicly available through the Digital Public Buyers Platform and the Living-in.EU website.
Support: This peer review exercise will be conducted with the support of Jeroen Naves, a lawyer specialising in technology and the law, DG GROW’s Digital Public Buyers Platform and Living-in.eu.
- Procurement of AI
- Publication date
- Submitted by
- EC content