Overview
Welcome to the first HEARTS workshop @CHI 2026! Rapid adoption of generative AI has outpaced the infrastructure
needed to red team systems responsibly. This workshop tackles a core tension: scaling AI red teaming while
centering human expertise and well-being. We convene academic, industry, government, and nonprofit practitioners to develop a shared vision for effective,
humane red teaming and to identify concrete ways to better support evaluators as red teaming scales.
We structure the workshop around two complementary threads:
- Vision: surface high-level goals and principles for scalable, accountable red teaming.
- Build: identify opportunities to support human–AI red teaming in practice (e.g., scenario libraries,
role prompts for red teamers, and calibration methods that align automation with human expertise).
Call for Participation
AI red teaming is increasingly central to identifying risks in deployed systems, yet it remains unclear how to scale
red teaming to meet organizational demands while protecting evaluator expertise and well-being. We welcome
participants from academia, industry, government, and nonprofits who work on AI evaluation, safety, and governance
including AI developers, AI red teamers, HCI researchers, policymakers, and community advocates.
Topics of interest include, but are not limited to:
- Automation levels & trade-offs: balancing scalability, coverage, validity, trustworthiness, and human oversight.
- Worker well-being & labor support: risk disclosure, rotation/recovery practices, training, and compensation structures.
- Tooling & infrastructure gaps: scenario libraries, annotation/triage/adjudication workflows, dashboards, and collaborative supports.
- Governance & reporting: protocols, documentation, and participatory safeguards that make red teaming accountable.
Options for Applying to the Workshop:
To ensure accessibility for all practitioners, we are providing 3 options to indicate your interest in our workshop:
- Answer short-form questions – ideal for practitioners who haven't written research papers (e.g., red teamers)
- Submit a statement of interest – ideal for academics who might not have relevant submissions or want to include specific information about your interests
- Submit a concurrent submission – ideal for those with accepted papers at relevant venues such as CHI, NeurIPS, ICML, CSCW, etc.
Please select only one of the options to complete.
Statement of Interest Guidance: For those who want to provide more details about your background and interest, we offer the option of submitting a statement of interest. The statement of interest should include information about relevant experience or interest you have to the workshop, what you hope to gain from attending, and what you hope to contribute to the workshop.
Review Process: The organizers will make selections prioritizing relevance and a diverse mix of perspectives. At least one author of each accepted submission must attend in person.
Templates:
Word
• LaTeX
• Overleaf
→ Submit
