Submission Guidelines

We invite extended abstract submissions of 2 pages (and additional pages for references, acknowledgments, and positionality statements) in ACM SIGIR format.

Submissions should be in English and not be anonymized (i.e., containing author names, affiliation, and e-mail address).

Statements should be submitted as PDF files via the EasyChair system by selecting the “2026 SIGIR Workshop on Vulnerabilities in Generative Systems for Information Retrieval” track.

Suitable LaTeX, Word, and Overleaf templates are available from the ACM Website (use sigconf proceedings template for LaTeX and the Interim Template for Word).

Submitting to this workshop indicates your agreement to and compliance with the common ACM and SIGIR 2026 Policies and Requirements.

Topics of Interest

We welcome a broad range of submissions, including visionary ideas, conceptual explorations, and preliminary empirical results. We strongly encourage submissions that offer theoretical perspectives or demonstrate practical system vulnerabilities.

Representative applications of generative systems include, but not limited to:

• LLMs for information seeking (QA systems, chatbots, etc.)
• LLM-as-a-judge (for evaluation, annotation, etc.)
• LLM for ranking (rerankers, etc.)
• LLM for query augmentation (generated query variants, etc.)
• LLM for simulation (synthetic data, LLM personas, etc.)
• Agentic IR systems (agentic echosystem, autonomous task executor, etc.)
• Personalized LLMs (personalized assistants, etc.)
• LLM for generative media (kinship-based role-play, virtual avatars, etc.)
• (…. tell us more in your submissions)

Representative vulnerabilities (i.e., downsides) of generative systems include, but not limited to:

• Black-hat SEO (document content injection, query injection, etc.)
• Adversarial attacks (corpus positioning, prompt injection, backdoor attack, etc.)
• Systemic bias or consequences (LLM narcissism, output bias, etc.)
• Misleading or harmful output (misinformation, disinformation, polarized output, etc.)
• Human-related consequences (intellectual laziness, emotional displacement, etc.)
• (…. tell us more in your submissions)

Extended abstract themes include, but are not limited to:

• System Vulnerability Statements: Novel attack schemas and the identification of flaws or biases within current system architectures and applications.
• User-Centric Statements: Research on how vulnerabilities of generative systems influence user behavior and information perception, with an emphasis on negative impact.
• Resources or Practical Demonstrations: Introduction of evaluation benchmarks, platforms, or open-source tools for the community; real-world case studies, including demonstrations of state-of-the-art attacks or exploits.
• Perspective or Theoretical Statements: Identifying new research trajectories, raising awareness of specific vulnerabilities and lacking resources, or proposing novel methodologies for vulnerability research; frameworks and models for characterizing and formalizing vulnerabilities in generative systems.
• Surveys and Reviews: Comprehensive overviews of the current landscape of generative system vulnerabilities.
• Industrial Perspectives: Hypotheses and practices regarding the real-world vulnerabilities of generative systems in production.

Program Committee

• Dr Nalin Arachchilage — RMIT University, Australia
• Prof. Pablo Castells — Universidad Autónoma de Madrid, Spain
• Prof. Charles Clarke — University of Waterloo, Canada
• Dario Di Palma — Polytechnic University of Bari, Italy
• Juwon Kim — Amazon, United States
• Udita Patel — Amazon, United States
• Dr Yongli Ren — RMIT University, Australia
• Dean Mark Sanderson — RMIT University, Australia
• Dr Paul Thomas — Microsoft, United States

(More to be confirmed…)

Selection Process

The selection process will be single-blind. The Program Committee will evaluate submissions based on: (1) relevance to the workshop themes; (2) significance of statements; and (3) the potential to stimulate meaningful discussion. Authors of accepted submissions will be invited to present at the workshop.

The accepted extended abstract will be non-archival. We strongly encourage authors with accepted submissions to upload their work to arXiv, and we will link to them on the workshop website and promote them accordingly.

Important Dates

All key dates are by the end of day in Anywhere on Earth (AOE).

• Extended abstract submission - 27 April 2026
• Notification - 21 May 2026
• Camera-ready submission - 15 June 2026
• Workshop co-located with SIGIR´26 - 24 July 2026