SIG-AI-ACT: Special Interest Group on Translating the EU AI Act into Technical Requirements Bridging regulation and technical design for trustworthy AI.
Website: https://sig-ai-act.github.io/
About SIG-AI-ACT
SIG-TREQ is a transdisciplinary initiative aimed at translating the principles of the EU AI Act—such as fairness, transparency, privacy, and robustness—into technical requirements and design practices for AI systems. The SIG will be structured into modular subgroups, each dedicated to key topics—such as risk classification, transparency, human oversight, privacy, and data governance—working toward concrete outputs like guidelines, evaluation methods, and tooling.
Founded by researchers at TU Delft, SIG-AI-ACT seeks to operationalize abstract legal and ethical principles through actionable, value-driven specifications, targeting sensitive and high-impact domains.
Our mission
To bridge the gap between legal obligations and technical practice in high-risk AI development, by building tools, frameworks, and design methodologies grounded in the EU AI Act.
Objectives
- Interpret and categorize regulatory language based on system type and risk levels.
- Translate legal principles into actionable technical specifications, reusable patterns, and best practices.
- Foster alignment across academia, industry, and regulators for trustworthy AI.
What We Do
SIG-AI-ACT operates as a modular, collaborative network involving academic researchers, practitioners, legal experts, policy experts, and industry partners. Core activities include:
- Working groups bringing together experts for collaborations on topics, such as fairness, transparency, robustness, and privacy.
- Regular sessions, reading groups, and interdisciplinary workshops.
- Case studies in collaboration with institutions and industry partners (e.g. healthcare).
- Living technical specification that evolves with legal and technological shifts.
- Evaluation frameworks to measure values like fairness and transparency in practice.
- Governance tools for developers, auditors, and regulators.
- Public consultations to prototype and reflect real-world needs.
Why?
There is a pressing need to bridge the gap between the legal obligations set out in the EU AI Act and their practical implementation. SIG-TREQ is essential to this effort. By interpreting key provisions of the Act it translates them into actionable technical guidelines, system design principles, and supporting tools. This work is critical to enable both practitioners and researchers to implement, evaluate, and align their systems with the Act’s requirements—whether in deployed applications or academic contexts such as research prototypes, experimental studies, and benchmark development.
