Call for Papers
We welcome submissions on evaluating and defining memorization, measuring tool-mediated data leakage, and understanding the privacy risks and failures that emerge from isolated or multi-step agent behavior. We are especially interested in work that makes these risks measurable, including benchmarks, auditing tools, deployment studies, and legal or technical mitigation strategies.
Topics of Interest
Submitted work should target the following topics, as technical or position contributions.
- Privacy definitions for foundation models and agentic systems, including contextual, legal, social, and formal perspectives.
- Memorization, leakage, and privacy attacks on models, including extraction, inference, reconstruction, prompt injection, cross-session leakage, and multi-agent risks.
- Benchmarks, audits, and realistic evaluations of privacy risks, memorization, and safeguard.
- Mitigations and formal protections, including access control, unlearning, privacy-preserving tool use, differential privacy, and verification.
- Trade-offs, governance, and societal impacts of privacy-sensitive AI, including utility, safety, transparency, compliance, and downstream harms.
Submission Details
We invite short papers of up to 4 pages, excluding references and appendices. All submissions should follow the official NeurIPS 2026 paper formatting guidelines. Submissions must be formatted using the workshop's LaTex template. (It will be shared soon). Authors must adhere to the prescribed formatting requirements, including margins, font sizes, and page limits. Submissions that do not comply with the formatting or submission requirements may be rejected without review. The paper checklist is not required. Submitted work should be original and unpublished, although work available on preprint servers such as arXiv is permitted. We will enforce a strict desk rejection policy that follows NeurIPS policy. In particular, we will consider desk rejection for the following reasons:
- Not adhering to the NeurIPS paper template.
- Not anonymizing submitted work.
- Not adhering to the reciprocal reviewing agreement.
- Submitting work that violates the NeurIPS code of Ethics.
- Submitting work that contains hallucinated content and citations.
- Not documenting the use of LLMs for important, original, or non-standard tasks.
Concerning the last point, we expect authors to document their methodology clearly for upholding scientific rigor and transparency standards. In case LLMs or agents are used for important, original, or non-standard tasks, we ask you to report this in the Appendix or the main text of your paper. The use of spell checkers and grammar suggestions, aid for editing purposes, and basic code assistance do not need to be documented. Further, authors are responsible for the entire content of the paper, including all text, figures, and references. Therefore, while authors are welcome to use any tool they wish for preparing and writing the paper, they must ensure that all content is correct and original. For example, high-level instructions could potentially result in hallucinations when generating plots, risking scientific integrity. There have been many cases of hallucinated citations in literature review, which violates the NeurIPS Code of Conduct. It is the author’s responsibility to verify the tools are used in a scientifically responsible manner. Relatedly, as you are the one taking responsibility for the work, agents and LLMs cannot be authors. Finally, we note that attempts at prompt injections as well as other attempts to manipulate reviewing is strictly prohibited.
All accepted papers must be presented in person at the workshop. This is a non-archival venue, and there will be no formal proceedings.
Position Papers
In addition to regular research papers, we welcome the submission of position papers. Position papers present a well-reasoned perspective on an important challenge, opportunity, or future direction relevant to the workshop themes. Rather than reporting completed research, they argue for a viewpoint that can stimulate discussion, identify open problems, challenge existing assumptions, or propose new research agendas.
Position papers will be evaluated differently from research papers. The primary criterion is whether they present a compelling, well-supported, and timely argument that can foster constructive discussion within the workshop community. Reviewers will assess the quality of the reasoning and supporting evidence, rather than whether they agree with the authors' position. Thought-provoking and even controversial viewpoints are welcome, provided they are argued respectfully and supported by appropriate evidence and references.
Position papers should meet the same scholarly standards as research papers, including clear motivation, engagement with related work, and appropriate citations. Pure opinion pieces without sufficient justification or technical papers that merely advocate the authors' own method are generally not appropriate.
We encourage authors to:
- Clearly state their position in the title and abstract.
- Present evidence and reasoning supporting their claims.
- Discuss credible alternative viewpoints and explain why they disagree.
- Conclude with concrete recommendations, open research questions, or a call to action for the community.
Position papers may address any topic within the scope of the workshop, including emerging challenges, evaluation practices, theoretical perspectives, legal or ethical considerations, benchmark design, deployment practices, policy implications, or future research directions.
Accepted position papers will be presented alongside research papers during the workshop.
Submission Portal
All submissions must be made through OpenReview . The workshop uses OpenReview to manage submissions, peer review, and author discussions.
Before submitting a paper, every author and co-author must have an up-to-date OpenReview profile. Profiles should include current institutional affiliations and, where possible, a complete publication record to ensure accurate conflict-of-interest detection and reviewer assignment. You can create or update your profile by logging into OpenReview and selecting your profile from the account menu.
Authors are encouraged to create or update their OpenReview profiles well in advance of the submission deadline. Newly created accounts that are not associated with an institutional email address may require manual moderation, which can take up to two weeks before the profile becomes active.
During the review process, submitted papers, reviews, and author responses will only be visible to the assigned reviewers and workshop organizers. Reviews are not public during the review period.
For questions about creating an account, managing your profile, or using the submission system, please consult the OpenReview FAQ .
Important Dates
- Submission deadline: August 29, 2026, Anywhere on Earth (AoE)
- Notification of decisions: September 29, 2026, AoE
- Workshop date: To be confirmed