Abdallah El Ali, a researcher in the Distributed & Interactive Systems (DIS) group at CWI, actively participated in drafting the rules on disclosure of deepfakes and certain AI-generated text for deployers. His contribution focused specifically on the human-computer interaction perspective, emphasising the importance of making the nuances of human-AI collaboration visible within AI disclosures. This work connects closely to his broader research agenda as leader of the DIS group’s research area on trustworthy human-AI interaction, which covers AI disclosures, evaluation biases in large language models, and the design of transparent, human-centred AI interfaces.

The code of practice was drafted by six independent experts through a stakeholder process coordinated by the European AI Office. More than 180 participants were involved via working group meetings, interactive workshops, feedback rounds, and support channels. Prof. Natali Helberger of the University of Amsterdam, co-founder of the AI, Media and Democracy Lab, also contributed to the code. El Ali regularly collaborates with this lab.

The code provides providers of generative AI systems with rules for watermarking and detecting AI-generated and AI-manipulated content. For deployers, it contains guidelines and a set of icons that can be used to label AI-generated content, including deepfakes and certain AI-generated text. The code is intended to help providers and deployers prepare for the AI Act’s transparency rules, which will become strictly applicable from 2 August 2026.