The rapid sophistication of Artificial Intelligence (AI) technology is creating a widening literacy gap between the technical and legal professions and posing complex challenges for atrocity crimes prosecutions worldwide. Courts and legal practitioners across diverse jurisdictions are increasingly confronting AI-affected evidence without consistent frameworks to guide their decisions on its admissibility and reliability. To address this critical gap, this initiative proposes the development of an open-source, centralized AI Evidence Knowledge Hub. The Hub is designed to be the go-to resource for stakeholders to understand and compare how different legal systems are approaching the challenges of AI in the courtroom.

The AI Evidence Knowledge Hub will be an open-source repository of cases, procedural rules, and other relevant materials concerning the use of AI-affected evidence in legal proceedings around the world.

The scope of the Hub will focus on "AI-affected evidence," a term describing evidence that has been generated, modified, or surfaced using AI. This includes:

• AI generated evidence, such as deepfakes.

• AI modified evidence, such as using AI to improve the clarity of a non-AI image.

• AI surfaced evidence, which is information found (e.g., through scraping) or filtered (e.g., through object detection or facial recognition technology) using AI.

The Hub's primary goal is to empower a global community of judges, lawyers, human rights fact-finders, criminal investigators, and civil society organizations by providing them with accessible, comparative information.

The AI Evidence Knowledge Hub aims to reduce the lag time between rapid technological change and the adaptation of legal processes. By providing a centralized, comparative resource, the Hub will support the development of more consistent and equitable standards for handling AI-affected evidence globally.

The Knowledge Hub will be developed using a decentralized and collaborative methodology to ensure a comprehensive and diverse collection of information.

The Hub will primarily be populated with open-source court materials from domestic legal proceedings, but may also include:

• open-source court materials from international/regional legal proceedings;

• open-source materials from non-judicial/quasi-judicial institutional proceedings or investigations (for example, by UN treaty monitoring bodies) involving or relying upon AI-affected information;

• Applicable legislation or treaty provisions;

• Secondary materials, including but not limited to academic journals, advocacy resources, technical best practices/guidelines.