Dalle Molle Institute for Artificial Intelligence (IDSIA USI-SUPSI), and University of Florence

Promote global standards and unified guidelines for developing human-centered trustworthy AI

• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight

We are witnessing unregulated and uncontrolled adoption of AI in several critical areas, including cyber-physical systems whose undependable operation can severely impact lives of many people. There is urgency to agree on global frameworks putting together the ongoing efforts made by diverse stakeholders worldwide.

The widespread introduction of "opaque" (not interpretable/explainable) Machine Learning modules without appropriate risk assessment poses many risks within cyber-physical systems, especially in the field of intelligent transportation and autonomous vehicles, but even in military decision support systems.

Putting together diverse stakeholderd and decision-makers can significantly push towards prioritizing the adoption of trustworthy AI mechanisms such as explainability/interpretability supporting transparency, human-machine teaming and human oversight.

I am aware and actively involved in several IEEE societies, technical committee and task forces such as those related to IEEE CIS SHIELD Technical Committee (https://cis.ieee.org/committees/technical-committees/shield). Building upon the current state of the art in technical standards, policies and regulations is essential for the development of a global picture from which to build clear global trustworthy AI guidelines and new task forces with clear objectives to enforce them.

There is the need for cross-diclipline teams including academics, politicians, lawyers, etc., with appropriate background on AI governance and willingness to contribute to the development of global unified guidelines for adoption of trustworthy AI worldwide, addressing current challenges and mitigating AI-related risks.

I think the IEEE (https://www.ieee.org/) is currently not adequately represented, although it is the community of experts, mainly computer scientists and engineers, who have the most in-depth competence in AI system development and management, contributing to world standards. Their motto is "Advancing Technology for Humanity", which is a sinthesis of what AI development should focus on, while minimizing the risks of technology misuse.

Join development of white papers as a result/output of moderated discussions would be an effective mean for active engagement of diverse stakeholders.

The IEEE (Institute of Electrical and Electronics Engineers) provides foundational guidance through the "Ethically Aligned Design" framework, which promotes human rights, transparency, accountability, and ethics-by-design approaches. The IEEE 7000 series defines concrete, implementable standards: IEEE 7000: model process for integrating ethical considerations into system design and requirements engineering IEEE 7001: transparency of autonomous systems, including documentation and explainability requirements IEEE 7002: data privacy process, supporting privacy-by-design across the data lifecycle IEEE 7003: algorithmic bias considerations, including methods for identifying, measuring, and mitigating bias IEEE 7010: well-being metrics for AI, enabling assessment of societal and human impact The IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems promotes multi-stakeholder governance and provides guidelines and best practices. Key practices enabled by IEEE standards include value-based design (translating stakeholder values into requirements), traceability of ethical requirements, lifecycle governance (from design to deployment and monitoring), and measurable compliance through defined metrics. IEEE also supports certification efforts (e.g., CertifAIEd) to enable independent verification of ethically aligned AI systems. Overall, IEEE contributes to AI governance by making ethics actionable through technical standards, measurable criteria, and engineering processes.