A Toolkit for Compliance, a Toolkit for Justice: Drawing on Cross-sectoral Expertise to Develop a Pro-justice EU AI Act Toolkit

Existing AI ethics toolkits have significant gaps: The paper identifies that current toolkits suffer from limitations because they're either created by industry teams focused narrowly on their own needs or by academics producing abstract guidance disconnected from real-world development practices—suggesting a need for better bridges between these worlds.

Compliance alone is insufficient for responsible AI: The authors argue that toolkits designed merely to help companies meet EU AI Act requirements miss an opportunity to address "broader socio-ethical questions," implying that legal compliance and genuine justice/fairness in AI are not the same thing.

Cross-sectoral collaboration is essential but challenging: The toolkit was developed through partnership between UK academics and an Italian industry team, with the paper emphasizing that meaningful collaboration requires active "negotiation efforts"—suggesting that creating truly useful AI ethics resources demands difficult conversations between different sectors with different priorities.

Academic-industry partnerships need intentional design: Rather than having academics hand down guidance or industry self-regulate, the paper proposes that effective AI ethics resources require deliberate partnership structures where both perspectives shape the final toolkit from the ground up.

Process matters as much as product: The authors position their development process itself as a "blueprint" for others, suggesting that how organizations create AI ethics guidance—through inclusive collaboration—may be as important as what guidance they ultimately produce.