Answers are grounded in this paper and cite their sources.
As research software grows more complex—and as AI-generated code and software agents become part of everyday development—the security of research software and its supply chain have become pressing concerns. Threat modeling is the standard engineering approach for understanding and mitigating security risks, yet it remains difficult to apply in the research software environment, where projects combine scientific code, external dependencies, distributed infrastructure, and diverse organizational practices. Many research software engineers recognize that vulnerabilities exist but lack a structured way to reason about their systems’ threat surfaces. With support from the NSF Research on Research Security program, our team is convening an interactive threat-modeling workshop at the USRSE’26 Conference in October. The workshop’s goal is to capture practitioner experience, identify areas of consensus, and produce a community-informed view of the research software threat landscape. Attendees will collaboratively analyze representative research software systems and identify threats related to software supply chains, AI-assisted development, dependency management, and organizational practices. At SC, we will share the workshop readout, summarize the resulting threat model and key findings, and invite discussion from the broader HPC and research software communities. We hope this follow-on conversation will help validate, refine, and extend the workshop results so they reflect the broader research computing ecosystem.
No takes yet. Share an insight, caveat, or question.
Carver et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: