Collaborative session notes:

Open session notes

Description
We invite participants to join us for a demonstration, panel conversation, and open discussion on the opportunities and concerns surrounding autonomous agents visiting data repositories. This session explores the rapidly evolving role of autonomous agents in research data ecosystems and examines their impact when interacting directly with scientific data in trusted environments. Examples include agents that ‘visit’ data in place through secure compute workspaces, and agents embedded within repository interfaces that provide guidance, recommendations, or workflow automation. The discussion will highlight both the opportunities these agents create for accelerating data management and analysis workflows and the governance challenges that arise from their ability to reason and act independently.
The implementation of autonomous agents is moving at vastly different speeds across the research data landscape. AI-enabled tools and agent architectures are advancing rapidly, while repositories, governance models, and policy frameworks evolve at a much slower pace. The session will explore critical questions of readiness, resilience, and responsible deployment. Can the Research Data Alliance help bridge these differences in velocity? How can we accelerate the development of governance and policy mechanisms so that they keep pace with emerging technologies capable of visiting data, mediating access, and acting at the gate between users and repositories? At times, the contrast between slow-moving policy development and fast-moving innovation can feel like watching a glacier and a bullet travel side by side.
To address these issues, the session brings together data stewards, infrastructure architects, AI and agent specialists, and policymakers to explore how autonomous agents may reshape continuity of access across crisis, transition, and everyday research scenarios. Through demonstration, panel insights, and community discussion, participants will identify the opportunities, challenges, and potential RDA pathways for coordinated action.
Principal Objective
To assess whether autonomous agents that visit, mediate, or act upon data in place can enhance data continuity, quality, and research resilience, and to determine whether there is interest within the RDA community in forming a coordinated activity such as an Interest Group or Working Group on autonomous agents and data visitation.
Supporting Objectives
1. Examine the current ‘access gates’ and bottlenecks (technical, organizational, legal, and ethical) that inhibit continuity of data access in crisis or transition scenarios, and consider whether autonomous agents could mediate or reduce these barriers.
2. Explore how autonomous agents, such as intelligent negotiator agents, access brokers, or federated compute orchestrators, could support resilient and secure data visitation by enabling computation where the data reside rather than relying on data movement.
3. Identify the governance, ethical, legal, and technical safeguards needed when agents make decisions, request permissions, or act independently at points of access within repositories and trusted research environments.
4. Highlight concrete use-cases that can serve as testbeds for agent-mediated data visitation, such as health data spaces, crisis and disaster-response datasets, federated science platforms, and environments requiring HPC-enabled or burst-compute access.
5. Discuss potential next steps for the RDA community, including recommendations, shared guidance, or the formation of a new Interest Group or Working Group to develop community principles, best practices, or pilot demonstrations on agent-mediated data visitation.
6. Engage session participants through open discussion and virtual polling to gauge interest, identify priority areas, and assess whether the community supports continuing this work through a structured RDA activity.

90-minute agenda (based on the presenters)
1. Welcome and introductions (5 minutes)
Presented by: Natalie Meyers and Francis P. Crawley
2. Agent(s) demonstration (10 minutes)
Sanchita Bhattacharya: ‘The ImmPort Data Repository: Establishment as a Domain Node in a Data Mesh for AI-Assisted Immunology Research.’
3. Panel presentations (30 minutes – 5 minutes per speaker)
Presenters:
• Gnana Bharathy: AI Agent risk and/ responsible AI
• Lars Eklund: ‘The role of Agentic AI in the EU data spaces’
• Paty Buendia: ‘Agents visiting data: Accelerating data quality, management, and sharing with autonomous agents’
• Ronit Purian: ‘Bridging Policy and Practice: Can AI Agents help?
• Francis P. Crawley: ‘The ENVISAGE Principles and PILOT Principles for AI Governance: Herding autonomous agents’
• Luis Jacob Retanan: “Agentic AI Use Case in Cognitive Warfare”
4. Panel discussion and Q&A (30 minutes)
All panelists, moderated by: Seonyoung Kim
5. Post-session poll and next steps (10 minutes)
Discussion of interest in possible RDA IG or WG
Led by: Natalie Meyers, Francis P. Crawley
6. Closing remarks (5 minutes)
Presented by: Seonyoung Kim
7. Adjourn

• Unlocking Biomedical Data Potential: The DV4RDA Project and the FAIRlyz Platform for Enhanced Data Sharing, Analysis, and Community Engagement (https://docs.google.com/document/d/1F4n1f0Aq4tX3yOfTk1ds3JWx7q9Ma5Fg6OXI0NI5bdI/edit?tab=t.0)
• Buendia, P., Kim, S., Meyers, N., Crawley, F. P., Farrell, G., Purian, R., & RDA Artificial Intelligence & Data Visitation WG. (2025). Geographies of Trust: AI, Biomedicine, and the Next Era of Federated and Visiting Data Models (1.0). Zenodo. https://doi.org/10.5281/zenodo.16929066
• Prototype to Production | Kaggle (https://www.kaggle.com/whitepaper-prototype-to-production)
• Guidance for Ethics Committees Reviewing Artificial Intelligence and Data Visitation (https://www.rd-alliance.org/groups/artificial-intelligence-and-data-visitation-aidv-wg/outputs?output=165867)
• Guidance for Informed Consent in the context of Artificial Intelligence and Data Visitation (https://www.rd-alliance.org/groups/artificial-intelligence-and-data-visitation-aidv-wg/outputs?output=143365)
• AI Bill of Rights Recommendation (https://www.rd-alliance.org/groups/artificial-intelligence-and-data-visitation-aidv-wg/outputs?output=165995)
• Secure Processing Environments for Open Science: Proposing Legal Foundations (https://www.rd-alliance.org/groups/artificial-intelligence-and-data-visitation-aidv-wg/outputs?output=175108)
• AIDV-WG Shared Citation Library (https://www.rd-alliance.org/groups/artificial-intelligence-and-data-visitation-aidv-wg/outputs?output=165997)