Making AI explainable requires more than algorithmic transparency: it demands understanding who needs explanations and why. Without AI explainability, there can be no AI accountability. Unaccountable AI leads to automated injustice.

Explainable AI (XAI) is central to building responsible AI systems. Yet LLM-based agentic AI (systems that plan multi-step strategies, invoke external tools, and trigger cascading real-world consequences) fundamentally challenges existing explainability paradigms. What does explainability mean when AI systems no longer produce a single output, but instead unfold over time through multi-step actions? What new challenges do these settings introduce and how can we address them? As these systems scale from research prototypes to production deployments, the gap between what stakeholders need to understand and what current XAI provides grows dangerously wide.

We invite you to the 6th Human-Centered Explainable AI (HCXAI) workshop at CHI 2026 (one of the longest running and widely attended workshops at CHI, since 2021 over 450 participants joined from 21 countries). In this edition, researchers, practitioners, and policymakers from HCI, AI, design, social science, and policy will reimagine explainability for agentic systems—not just algorithmic transparency, but understanding who needs to know what, when, and why. 

Read the HCXAI 2026 workshop proposal by clicking here. 

14:15 | Intro & Welcome

14:35 | Lightning Talks I: Psychology, Communication, and the Design of XAI

• Less Interaction But More Explanation: A Communications Perspective on Agentic AI Interfaces Eunchae Jang; S. Shyam Sundar
• Not All Explanations Are Sought: Information-Seeking Psychology for Human-Centered XAI. Andrea Beretta and Salvatore Rinzivillo
• LLM Self-Explanations as Design Material: Toward a Taxonomy. Willem van der Maden, Pelin Karaturhan, Wendy Zhou and Jichen Zhu

14:55 | Lightning Talks II: Reasoning Made Legible: Developer Tools, Process Visibility, and the Limits of CoT

• Chain-of-Thought: Epistemic Flaws and Fictional Explanations. Fabio Morreale, Joan Serrà and Yuki Mitsufuji
• Beyond Output Explanations: Process-Level Legibility for Agentic AI Systems. Ning Coeva
• From Black Box to Toolbox: Human-Centered Explainability in Developer-Authored Agent Rules. Lilia Pérez Romero, Tao Dong and Ben Ferrari-Church

15:15 | Lightning Talks III: Building for Everyone: Stakeholder-Differentiated Approaches to XAI at Scale

• Agentic Explainability at Scale: Between Corporate Fears and XAI Needs. Yomna Elsayed and Cecily Jones
• Beyond the 'Diff': Addressing Agentic Entropy in Agentic Software Development. Matteo Casserini, Alessandro Facchini and Andrea Ferrario
• The interdisciplinary infrastructure of explainability: XAI stakeholders, tools and approaches: two empirical case studies from the FinTech industry. Wojtek Buczynski and Jingkun Zhu
• Not One Size Fits All: A Stakeholder-Differentiated Framework for Explainable Agentic AI in Education. Judith Kankam-Boateng

15:45 | BREAK ☕️

16:30 | Poster Spotlight I Explanations in Communicative/Design Artifacts

• Reframing Explainability in Agentic AI Systems:Temporal Openness, Narrative Reversibility, and the Civilizational Function of Interpretation. Lyric Li
• AI-Driven Service Blueprints As a New Methodological Tool for HCAI Experience Design. Mehrdad Atariani and Oluwole Ajala 
• “Can Make Mistakes”: AI Chatbot Disclaimers as Failed Explainability Surfaces. Sasha Mitts
• AI Explanations as Boundary Objects: Toward Shared Language in High-Stakes Decision-Making. Kira Clements

16:40 | Poster Spotlight II Explainability Across Ability, Language, and Clinical Context

• The Medicalization Gap in Accessible Explainable AI. Retno Larasati and Alaa Hammad
• Designing Explainable Conversational Agentic Systems for Guaraní Speakers. Samantha Adorno, Akshata Kishore Moharir and Ratna Kandala
• High-Stakes Clinical AI Teaming, Not Transparency: Explainability as Coordination Infrastructure. Dezhi Wu
• Explainable AI for Blind and Low-Vision Users: Navigating Trust, Modality, and Interpretability in the Agentic Era.  Abu Noman Md Sakib, Protik Dey, Zijie Zhang and Taslima Akter

16:50 | Poster Spotlight III Action, Inference, and Impasse: XAI in Agentic Contexts

•  Explainable Model Routing for Agentic Workflows. Mika Okamoto, Ansel Erol and Mark Riedl.
• Beyond Explanation: Deadlock Situations as a Stress Test for Human-Centered XAI in Automated Vehicles. Cansu Demir
• Counterfactual Explanations for Agentic Workflows. Madhuri Singh, Grace Kim, Mika Okamoto, Aarushi Ammavajjala, Amal Alabdulkarim, Gennie Mansi and Mark Riedl

17:00 | Poster Spotlight IVX AI in Detection, Moderation, and Algorithmic Oversight

• Understanding the Role of Visual Explanations in Human-AI Collaborations in Deepfake Image Detection. Min Zhang, Soraya Kouadri, Patrick Wong, Mark McJury, Nitu Bharati and Arosha Bandara
• Position Paper: Explainable AI (XAI) in Financial Crimes Detection. Guanming Shi and Yixin Zou
• Algorithmic Mirror: Interactive Visualizations for Adolescents to Understand and Challenge Algorithmic Profiling in Online Platforms. Yui Kondo, Kevin Dunnell, Isobel Voysey, Qing Hu, Victoria Paesano, Phi H Nguyen, Qing Xiao, Jun Zhao and Luc Rocher
• From Black-Box Filters to Agentic Pipelines: Designing Calibrated Reliance in Video Moderation. David Sarlos

17:10 | Group Activity
17:50 | Closing Remarks

Making AI explainable requires more than algorithmic transparency: it demands understanding who needs explanations and why. In our sixth CHI workshop on Human-Centered XAI (HCXAI), we shift focus to agentic AI systems. LLM-based agents foundationally challenge existing explainability paradigms. Unlike traditional AI that produces single outputs, agents plan multi-step strategies, invoke tools with real-world consequences, and coordinate with other systems; yet current XAI approaches fail to address these complexities. Users need to understand not just what an agent might do, but the cascade of actions it could trigger, the risks involved, and why responses take time. Even our expanded HCXAI frameworks struggle with these new demands. Through our workshop series, we have built a strong community making important conceptual, methodological, and technical impact. This year, we re-examine what human-centered explainable AI means in the agentic era, practitioners to shape explain- and developers of these systems.

Topics of Interest

We welcome position papers, empirical studies, system prototypes, critical reflections, design fictions, and system demonstrations across four interconnected themes:

1. Explainability Needs & Stakeholder Perspectives

• What users and developers need to know about agent behavior, before, during, and after execution?
• Bridging gaps between technical transparency and genuine understanding across stakeholders (end users, domain experts, developers, auditors, policymakers)
• Domain-specific needs in high-stakes contexts (healthcare, justice, education, hiring)

2. Explaining Agentic Behavior: Plans, Tools & Traces

• Multi-step plans, tool invocations, and cascading real-world consequences
• The limits of chain-of-thought: are they actual explanations?
• Excusable vs. explainable AI: distinguishing between justifying decisions and enabling genuine understanding

3. Trust, Accountability & When Explanations Fail

• Calibrated reliance: appropriate trust vs. dangerous over-reliance
• XAI as accountability infrastructure: legal, regulatory, and governance dimensions
• Dark patterns & failure modes: misleading explanations, explanation fatigue, cognitive overload, manipulation
• Autonomy vs. accountability: The relationship between agent autonomy and human accountability

4. Evaluation, Participation & Futures

• Sociotechnical benchmarks: evaluation beyond technical metrics, measuring real understanding and decision quality
• Participatory and co-design approaches centering affected communities
• Provocations & speculative futures: What should agentic Human-centered XAI look like in 2030?

We especially welcome work that:

🔥 Challenges assumptions about what constitutes explainability
🚨 Exposes limits, failures, and unintended consequences
🌍 Bridges disciplines—HCI, AI, social science, law, design, domain expertise
🛠️ Proposes novel interaction paradigms or evaluation methods

No single discipline, method, or perspective is privileged. We welcome multidisciplinary work. 

Authors are invited to submit papers 2-5 (FIVE) pages excluding references. Papers should be formatted in accordance with the single-column ACM SIGCHI format. Online guidance is available: https://www.acm.org/publications/authors/submissions. 
 

Templates are available for the following platforms: 

• Overleaf (Latex) 
• Microsoft Word
• LaTeX  (Use sample-manuscript.tex for submissions, and use \documentclass[manuscript, review]{acmart}.)

Reviewers will review the papers in the single-column format. Contact authors of accepted papers will receive instructions on how to prepare and submit a final version by the Publication-Ready Deadline. 

Submissions are single-blind reviewed; i.e., submissions must include the author’s names and affiliation. The workshop's organizing and program committees will review the submissions and accepted papers will be presented at the workshop. We ask that at least one of the authors of each accepted position paper attends the workshop. Presenting authors must register for the workshop and at least one full day of the conference.

Submissions must be original and relevant contributions to the workshop's theme. Each paper should directly and explicitly address how it speaks to the workshop's goals and themes. Pro-tip: direct mapping to a question or workshop goal posed will help. Examples include, but not limited to, papers that include research summaries, literature reviews, industrial perspectives, real-world approaches, study results, or work-in-progress research projects. At the very least, submissions will be hosted on the website in line with what we have done for past years. 

Submission pro tips:

1. Explicitly align your submission with the workshop's goals and topics. How? (a) Refer to the questions in the Call for Papers. (b) Read the workshop proposal

2. Engage with past submissions (build on, don't repeat). This year, we are putting extra emphasis on how authors are building on prior papers in this workshop. All papers are available on the website. Please engage with them, and build on them. 

3. Provocations/Critical Reflection Position papers must make a well-justified argument, not just summarize findings. This means that even if you are summarizing findings, make an argument around that summarization and justify why that argument (position) is something that is discussion-worthy and valuable to the community.
4. Research/empirical/system papers must provide well-articulated contributions, not just imagined studies. Preliminary studies and results are fine. Imagined/proposed studies are not. Research papers should outline a clear problem that is addressed by a study (or meta-analysis or reanalysis of previous studies). The contributions must be clear and well-situated in the literature. 

We aim to have global and diverse participation in the workshop given its hybrid (virtual-first) design format reduces visa or travel-related burdens,. With an effort towards equitable conversations, we welcome participation from under-represented perspectives and communities in XAI (e.g., lessons from the Global South, civil liberties and human rights perspectives, etc.)

Submit Your paper (On EasyChair)