Thank you to our sponsors!
Colocated with the 2026 Annual Conference of the Association for Computational Linguistics (ACL 2026)
Room: Harbor G, Manchester Grand Hyatt San Diego, California, United States
With the rapid advances in AI, empowered by large language models (LLMs) and natural language processing (NLP) techniques, there is an increasing integration of AI systems that directly interact with users and facilitate our daily tasks. In particular, the recent development of agentic models allows users to communicate directly with AI for complex tasks such as coding, web surfing, information seeking, and deep research. These models integrate NLP techniques with computer vision, systems engineering, and other social and physical sciences, expanding the boundaries of what AI systems can accomplish and making NLP systems omnipresent in various aspects of our everyday life. This makes the development of reliable, responsible, ethical, and safe AI increasingly important.
This year, we are excited to host our TrustNLP workshop at ACL 2026, inviting participants and papers that focus on developing models that are explainable, fair, privacy‑preserving, causal, and robust. In particular, we have secured sponsorship from major companies in the field, including Meta, Capital One, and Amazon. We will use the funding to promote diversity, participation, and mentoring, furthering our mission.
Invited Talk 1
Michael Johnston — Applied Science Manager, Responsible AI, Amazon AGI
July 4, 2026, 9:05 AM — Room: Harbor G
As AI systems become increasingly capable, it is critical that techniques ensuring their safety and alignment to human values keep up with the exponential pace of innovation. At the same time, there is increasing concern from an evaluation science perspective that static benchmarks fail to capture the true capabilities of models with respect to both utility and safety. Also, while high quality diverse data is critical for training, it is rare and expensive to create, especially for new tasks and multi-turn conversations. In this talk, I will illustrate how these three issues can be addressed through an 'Adversarial Arena' approach to driving research and data creation, where different approaches are evaluated through interactive competition. I will draw on examples and learnings from the Amazon Nova AI Challenge: Trusted AI, an international AI competition now in its second year. In the challenge, competing teams build either secure coding agents or automated red teaming bots and their creations face off in a series of tournaments setting in motion a continuous flywheel of innovation and data generation.
Michael Johnston is Applied Science Manager in the Responsible AI team in Amazon AGI. Michael has over 30 years of experience in artificial intelligence and machine learning and research contributions spanning NLP, dialog, multimodality, fusion of human and artificial intelligence, and trustworthy AI. Before joining Amazon, he was VP of Research and Innovation at Interactions Corporation, and earlier held positions at AT&T Labs Research, Oregon Graduate Institute, Brandeis University, and Apple. Michael has over 60 U.S. patents, and has published over 80 scientific papers. He has designed and overseen multiple international challenge competitions in artificial intelligence, including the Alexa Prize, the Amazon Trusted AI Challenge, and Amazon Nova AI Challenge: Trusted Software Agents.
Invited Talk 2
Lilly Weng — UC San Diego
July 4, 2026, 11:00 AM — Room: Harbor G
In this talk, I will present recent work from my lab toward trustworthy language models through representation-level interpretability, behavior steering, and principled evaluation. In particular, I will discuss: (i) recent findings showing that behaviors such as self-reflection and reasoning dynamics are encoded in the internal representations of language models and can be manipulated to steer model behavior and improve reasoning efficiency; (ii) emerging frameworks for rigorously evaluating the faithfulness of neuron- and representation-level explanations; and (iii) recent efforts toward trustworthy reasoning and interpretable-by-design language models, including training-free model editing, structured reasoning supervision, and architectures with explicit concept bottlenecks. These works illustrate a broader perspective for trustworthy language models: achieving reliable language models requires not only strong capabilities, but also principled mechanisms for understanding, evaluating, and controlling their internal behaviors.
Lily Weng is an Assistant Professor in the Halıcıoğlu Data Science Institute at UC San Diego and she leads the Trustworthy Machine Learning Lab at UC San Diego. She received her PhD in Electrical Engineering and Computer Science (EECS) from MIT in August 2020, and her Bachelor and Master degree both in Electrical Engineering at National Taiwan University. Prior to UCSD, she spent 1 year in MIT-IBM Watson AI Lab and several research internships in Google DeepMind, IBM Research and Mitsubishi Electric Research Lab. Her research interest is in machine learning and deep learning, with primary focus on Trustworthy AI. Her vision is to make the next generation AI systems and deep learning algorithms more robust, reliable, explainable, trustworthy and safer. Her work has been recognized and supported by multiple NSF awards, ARL award, Intel Rising Star Faculty Award, Hellman Fellowship, and Nvidia Academic award. For more details, please see https://lilywenglab.github.io/.
We invite papers that focus on different aspects of safe and trustworthy language modeling. Topics of interest include (but are not limited to):
We welcome contributions that also draw upon interdisciplinary knowledge to advance Trustworthy NLP. This may include working with, synthesizing, or incorporating knowledge across expertise, sociopolitical systems, cultures, or norms.
All submissions undergo double‑blind peer review (with author names and affiliations removed) by the program committee, and they will be assessed based on their relevance to the workshop themes.
All standard submissions go through the OpenReview platform. To submit, use this submission link.
Submitted manuscripts must be 8 pages long for full papers and 4 pages long for short papers. Please follow ACL submission policies. Both full and short papers can have unlimited pages for references and appendices. Please note that at least one of the authors of each accepted paper must register for the workshop and present the paper.
Template files can be found here.
We also ask authors to include a limitation section and broader impact statement, following guidelines from the main conference.
If your paper has been reviewed by ACL, EMNLP, EACL, or ARR and the average rating is higher than 2.75 (either average soundness or excitement score), the paper is qualified to be submitted on the fast track. In the appendix, please include the reviews and a short statement discussing what parts of the paper have been revised.
Fast-Track submissions go through the OpenReview platform. To submit, use this submission link.
ACL workshops are traditionally archival. To allow dual submission of work, we are also including a non‑archival track. If accepted, these submissions will still participate and present their work in the workshop. A reference to the paper will be hosted on the workshop website (if desired), but will not be included in the official proceedings. Please submit through OpenReview but indicate that this is a cross‑submission at the bottom of the submission form. You can also skip this step and inform us of your non‑archival preference after the reviews. Papers accepted to the Findings of ACL 2026 may also submit non‑archival to the workshop.
Accepted and under‑review papers are allowed to be submitted to the workshop but will not be included in the proceedings.
No anonymity period will be required for papers submitted to the workshop, per the latest updates to the ACL anonymity policy. However, submissions must still remain fully anonymized.
Interested in reviewing for future editions of TrustNLP?
Please fill out this form.
Please contact us at trustnlpworkshoporganizers@gmail.com.