Paper
Anonymous submissions limited to 9 pages (excluding references), using ACM Conference Proceeding templates; supplemental material (code, proofs, pseudo-code) allowed with no page limit but not required for review.
AI/CORE Unranked
KDD 2026
KDD workshop on Evaluation and Trustworthiness of Agentic AI
The KDD 2026 Workshop on Evaluation and Trustworthiness of Agentic AI focuses on developing advanced methodologies to assess and ensure the reliability, safety, and governance of autonomous AI agents in real-world, dynamic environments. It addresses challenges such as post-market monitoring, model evolution, multi-agent interactions, and the lack of standardized benchmarks, bringing together researchers and industry practitioners to advance evaluation frameworks beyond traditional static testing.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Research areas in scope
01
Topics of Interest
Real-Time Post-Market MonitoringAgentic AI EvaluationModel Evolution and API RiskTrustworthiness and SafetyBenchmarking, Metrics, and StandardizationLifecycle and Governance FrameworksUser-Centric and Cross-Modal AssessmentIndustrial and Public-Sector Applications
Policies worth checking twice
- Paper submissions must be anonymous.
- Papers are limited to 9 pages excluding references.
- Submissions must be in PDF format using ACM Conference Proceeding templates.
- Supplemental material for reproducibility is allowed with no page limit but is not required for review.
- Accepted papers will be posted on the workshop website but will not be included in the KDD proceedings.
- Submissions must be made through OpenReview.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026