The final day to reserve a room in our block is October 18th. As we sold out last year, we advise you to book your hotel room as early as possible.

The global focus of this symposium track is trustworthy, safe, and assured AI across the full AI system lifecycle. AI assurance must address both individual AI components and the larger sociotechnical systems in which they are embedded. The track emphasizes methods that provide evidence, metrics, and bounds for fairness, reliability, robustness, security, privacy, reproducibility, accountability, transparency, and resilience in the context of quantifying and managing AI-system risk. It spans foundational research, metrology, formal methods, systems engineering, software engineering, human factors, policy, governance, and operational practice. The symposium will bring together researchers and practitioners from academia, industry, government, standards organizations, and civil society who are invested in addressing the scientific and engineering challenges of AI assurance in applications where a priori understanding of risk is critical.

Topics of interest include, but are not limited to:

• Trustworthy multi-agent and agentic AI systems, with emphasis on governance, alignment, robustness, reliability, controllability, accountability, delegation, and emergent behavior in risk-averse contexts.

• Sociotechnical perspectives on AI-enabled systems, including the relationship among technical performance, human trust, organizational accountability, governance, standards, metrology, and social-science and humanities methods.

• Approaches for enhancing reasoning in large language, foundation, and multimodal models, including causal reasoning, neuro-symbolic methods, retrieval-augmented generation, formal reasoning, outcome verification, and process supervision.

• Verification, validation, and testing of AI systems, including quantitative AI and system performance indicators, operational design domain specification, confidence bounds, and links among performance, trustworthiness, and trust.

• Evaluation of AI system vulnerabilities, risks, and impacts, including prompt injection, data poisoning, model extraction, tool misuse, adversarial manipulation, automated red teaming, degradation objectives, and liability assessment.

• Governance, standards, policy, and regulatory approaches for trustworthy AI, including AI management systems, conformity assessment, risk classification, independent evaluation, and cross-border interoperability.

• Neuro-symbolic, causal, and knowledge-based methods that combine data-driven learning with domain knowledge to support reliability requirements, quantify uncertainty, reduce overgeneralization, and improve the trustworthiness of AI-enabled critical applications.

For more information on topics, see our Call for Papers page.