The AI Office is hosting an online workshop (tentatively 15 July 2026) focusing on independence and qualification requirements that organisations external to GPAI model providers should meet for the purposes of conducting model evaluations and assessing systemic risks in accordance with the GPAI Code of Practice, particularly when GPAI model providers engage with external evaluators.
The event will bring together AI safety experts and AI Office staff to exchange views on independence and qualification standards that external evaluators should satisfy when engaged by GPAI model providers to evaluate GPAI models and assess their systemic risks. The workshop will gather expert opinions to inform the development of robust requirements ensuring the credibility and reliability of external evaluations under the AI Act.
Eligibility and selection
The AI Office individuals and organisations with relevant expertise to express their interest in participating, by 21 June 2026.
Following the selection procedure, invitations will be sent by 7 July 2026.
Applicants will be selected based on:
- Demonstrated experience in conducting high-quality evaluations of GPAI models and their systemic risks (e.g. in-depth evaluations of capabilities, propensities, risks, safety mitigation effectiveness, including red teaming, uplift studies, white-box evaluations) and/or
- Quality of the track record of AI governance thought leadership with a focus on evaluations of systemic risks (e.g. prominent white papers).
Individuals or organisations that already belong to the European AI Office's Network of Evaluators and can demonstrate a targeted focus on the topic of the workshop only need to indicate their membership in the application. (An individual or organisation is part of the Network of Evaluators if they previously gave such consent when participating in either workshops on 13 December 2024 or 28 April 2025.)
Background
The AI Act establishes rules to ensure general-purpose AI models are safe and trustworthy, particularly those posing systemic risks. The GPAI Code of Practice sets out how providers can demonstrate compliance, including through external evaluations. Ensuring that external evaluators meet appropriate independence and qualification requirements is critical for the credibility and effectiveness of this evaluation framework. The AI Office is engaging with the expert community to develop these requirements.
Disclaimer
The workshop is not intended to cover the expertise, competence, independence or any other requirements established by the AI Act in the context of the scientific panel (Article 68(2) of the AI Act or the independent experts that the Commission may appoint to conduct evaluations on the basis of Article 92(3) of the AI Act.