The UNIFAI Toolkit is a comprehensive open-source software platform designed to support organizations, researchers, developers, and policymakers in conducting Algorithmic Impact Assessments (AIIAs). It was developed collaboratively by AiNTIGONE, the Distributed Systems Laboratory (DS Lab) at the National Technical University of Athens (NTUA), and the Data & Cloud Research Group (DAC) at the University of Piraeus (UniPi). The toolkit operationalizes core principles of ethical and trustworthy AI, drawing on established and emerging regulatory frameworks including the EU Artificial Intelligence Act, Data Act, Data Governance Act, and the ISO 42001 standard for Art
The platform is built upon a modular and extensible architecture, reflecting a lifecycle-oriented design inspired by ISO 42005 (AI System Impact Assessment). It is organized into clearly defined sections that guide users through essential elements such as: AI system description and purpose (Section A), detailed dataset analysis and quality indicators (Section B), algorithm/model origin and development processes (Section C), deployment environments (Section D), identification of relevant stakeholders (Section E), assessment of actual and potential benefits and harms (Section F), and documentation of misuse scenarios and failure risks (Section G). Each section is accompanied by guidance prompts and documentation templates, enabling both AI producers and consumers to meet compliance obligations and contribute to ethical risk mitigation. The toolkit is particularly suited for high-risk AI systems, helping organizations preemptively identify problematic use cases, biases, or accountability gaps before deployment.
Released under the MIT License, the UNIFAI Toolkit is freely available for modification and integration into organizational workflows or AI governance platforms. It is designed to be a living, evolving system, allowing contributions from a broad community—including auditors, conformity assessment bodies, regulatory consultants, researchers, and domain experts. Its collaborative development model encourages updates based on policy changes, use-case feedback, and new AI paradigms such as neurosymbolic AI, active learning, or swarm intelligence. A live demonstrator instance of the toolkit is available at aia.dslab.ece.ntua.gr, providing users with hands-on access to its features, templates, and assessment workflow. By combining transparency, traceability, and a commitment to open science, the UNIFAI Toolkit sets a new standard for aligning AI development with societal values, ethical norms, and legal mandates.