Call for Papers
We invite submissions to the first ICML Workshop on Technical AI Governance (TAIG), which will be held at the Vancouver Convention Center on the 18th or 19th of July.
We invite submissions to the first ICML Workshop on Technical AI Governance (TAIG), which will be held at the Vancouver Convention Center on the 18th or 19th of July.
The rapid advancement of artificial intelligence systems presents unprecedented challenges and opportunities for governance. This workshop aims to bring together researchers from technical and policy backgrounds to address the growing need for effective governance frameworks for AI systems. We seek contributions that bridge the gap between technical capabilities and governance requirements.
We welcome submissions in all areas of Technical AI Governance, including but not limited to (see Reuel et al., 2024 for a more comprehensive overview of topics):
Evaluation and benchmarking methodologies for AI systems [Piktus et al., 2023; Perez et al., 2023; Casper et al., 2024; Eriksson et al., 2025]
Methods for facilitating secure external evaluations of AI systems [Beer and Toner, 2025; Ojewale et al., 2025; South et al., 2024]
Verification of AI system properties [Choi et al., 2023; Katz et al., 2017; Jia et al., 2021]
Robustness and security in AI [Hubinger et al, 2024; Nevo et al 2024]
Transparency and documentation standards [Mitchell et al., 2019; Zhang et al., 2024; Bommasani et al., 2023]
Hardware-based mechanisms for secure and confidential AI [Mo et al., 2024; Li et al., 2023]
Analysis of industry trends [Sevilla et al., 2022]
We also welcome less technical papers that address topics regarding TAIG as a field, including:
Surveys and reviews [Buhl et al., 2025; Robinson et al., 2025]
Governance implications of recent technical developments in AI e.g. reasoning models, agentic systems, scaling test-time compute [Toner et al., 2025; Chan et al., 2025]
Technical analyses of existing or proposed policies [Hooker, 2024; Marino et al., 2025]
Position papers, e.g. Challenges and limitations of technical AI governance [Reuel et al., 2024b]
Submissions will be made through the OpenReview portal, available here.
Format
Submissions must use the ICML 2025 template (available here) and must not exceed 4 pages, excluding references and appendices which may be unlimited in length (although reviewers are not required to read appendices). All submissions must be anonymous to accommodate our double-blind review process. Any submissions that do not meet the anonymity, page limit, or formatting criteria will be rejected without review.
Selection Criteria
We encourage interdisciplinary submissions that are well-situated in the current discourse regarding both technical and policy considerations of AI. This means that authors of more technical-leaning papers should meaningfully interact with the policy motivations and implications of their work, and, where suitable, provide specific policy recommendations. Similarly, more policy-focussed submissions should engage substantively with the technical aspects of AI research, development, and use. Our review process will reflect this interdisciplinary nature, with papers evaluated by reviewers with expertise in both technical and policy aspects of AI governance.
Fast-Track Submissions
We welcome submissions of papers that have already undergone peer review at other venues. There will be an option to request fast-track in the submission portal. If selected, you will be prompted to provide evidence and details of previous peer review (e.g. a link to publicly visible reviews on OpenReview). Note that fast-track submissions are subject to the formatting requirements detailed above, and that prior acceptance to another venue does not guarantee acceptance to this workshop – fast-track submissions will be considered by the program committee alongside all other submissions once reviewing has been completed.
Publication and Presentation
The workshop is non-archival, and accepted papers will be published on the workshop website. All accepted papers will be presented as posters during the workshop, for which at least one paper author must be present in-person. Additionally, a select number of outstanding papers will be invited for lightning talks.
Important Dates
Submissions Open: Wednesday, April 23rd
Abstract Submission Deadline: Wednesday, May 7th (11:59 PM AOE)
Full Paper Submission Deadline: Monday, May 12th (11:59 PM AOE)
Acceptance Notification: Wednesday, June 4th
Camera-Ready Deadline: Wednesday, July 2nd
Workshop Date: Friday/Saturday, July 18th/19th
Contact
In case of any questions, please contact the workshop organisers at hi@taig-icml.com
Yes. Our workshop has non-archival proceedings.
Yes. We encourage submissions of ICML 2025–accepted papers that align with the workshop’s theme of technical AI governance. If your paper has already been peer reviewed, you are welcome to use our fast-track option. Acceptance to ICML 2025 does not guarantee acceptance to this workshop.
Yes. Concurrent or subsequent submissions to other venues are generally acceptable given our non-archival policy.
Yes. However, there will be no rebuttal phase.
Yes. At least one author should attend and present the work (poster session by default).
Please email the workshop organisers at hi@taig-icml.com.