Overview
We invite original, unpublished research on secure, explainable and robust governance of AI-generated content. AIGCTrustDetect goes beyond binary detection toward full-stack governance, and especially welcomes work with clear potential for generalization and cost-effective real-world deployment.
Accepted papers will be presented at the workshop, held as part of COLING 2027 in Macau.
Topics of Interest
Topics include, but are not limited to:
- Reliable and explainable detection. Detection methods for AI-generated text: zero-shot and supervised detectors, watermarking and provenance signals, stylometric and statistical features; interpretability, calibration and uncertainty estimation; attribution to specific models or generators; evaluation protocols, benchmarking and error analysis; decisions that humans can verify and contest.
- Detection across modalities. AI-generated content detection beyond text: speech, image, video and multimodal content, including deepfakes and synthetic media, and the transfer of detection techniques across modalities.
- Detection in complex generation scenarios. Mixed human–AI authorship and LLM-refined text; paraphrasing, translation and other obfuscation; adversarial attacks on detectors and deliberate evasion; robustness and generalization across models, domains, prompts and decoding strategies; real-time and large-scale detection systems.
- Detection and governance in agentic workflows. Content produced by LLM agents and tool-using systems: provenance and audit trails across multi-step generation, detection and attribution of AI-generated code, reliability of agent-written reports, and human oversight of automated content pipelines.
- Cross-lingual generalization. Multilingual and cross-lingual detection; transfer and low-cost adaptation to low-resource languages; language-agnostic signals; multilingual datasets and benchmarks.
- Linguistic perspectives on AI-generated text. Syntactic, semantic, discourse and pragmatic signatures of machine writing; corpus-based and stylometric comparison of human and machine text; register, genre and stylistic variation across languages; linguistic typology and its implications for detection; and human perception of AI-generated language.
- Hallucination, reliability assessment and factuality verification. Whether generated content is accurate, supported and proportionate: hallucination detection, evidence grounding and claim verification, evidence–statement mismatch and overclaim, fake news and misinformation, reliability assessment of generated text, and grounding and citation fidelity in retrieval-augmented generation.
- AI assistance in scholarly writing and peer review. Detection and disclosure of AI-generated manuscripts and reviews, reliability of LLM-as-a-judge evaluation, and governance of AI assistance across the research pipeline.
- Hierarchical and differentiated governance. Governance itself as a research problem: tiered frameworks and new tasks that formalize proportionate response across settings with different stakes, from education to journalism to content platforms; disclosure, provenance standards and policy; fairness and disparate impact of detectors, safeguards against false accusation; ethical, legal and societal implications of AI-generated content.
- Synthetic data and training-set governance. Detection and labeling of AI-generated content in training corpora, reliability of synthetic data, contamination and deduplication, and implications of recursive training for future detectors.
- Datasets, resources and tooling. Corpora, benchmarks, evaluation tools and open-source infrastructure supporting any of the directions above.
Where Does Your Paper Fit?
Select everything your work does and watch the stack light up. This is an informal guide for authors; it plays no part in reviewing.
My work…
Submission
We invite long and short papers, with the same page limits as the COLING 2027 main conference, formatted in the COLING 2027 style [template link TBD]. Papers can be submitted directly through OpenReview [link TBD], or committed with existing reviews and meta-reviews from ACL Rolling Review (ARR) through the OpenReview commitment page [link TBD]. All submissions are reviewed double-blind, and each paper receives at least two reviews.
We cannot accept submissions for publication that will be (or have been) published elsewhere. All accepted submissions will be presented as talks and/or posters at the workshop, and will be included in the workshop proceedings.
Important Dates
| First call for papers | Friday | October 30, 2026 |
|---|---|---|
| Second call for papers | Wednesday | November 25, 2026 |
| Direct paper submission deadline | Tuesday | January 5, 2027 |
| Pre-reviewed ARR commitment deadline | — | TBA |
| Notification of acceptance | Saturday | February 6, 2027 |
| Shared task call for participation | Tuesday | November 10, 2026 |
|---|---|---|
| Registration opens | Tuesday | November 10, 2026 |
| Training and development data release | Wednesday | November 25, 2026 |
| Test data release | Sunday | December 20, 2026 |
| System output submission deadline | Friday | December 25, 2026 |
| Evaluation results release | Wednesday | December 30, 2026 |
| System description paper deadline | Sunday | January 10, 2027 |
Common Dates
| Camera-ready deadline | Friday | March 5, 2027 |
|---|---|---|
| Proceedings due (hard deadline) | Monday | April 5, 2027 |
| Pre-recorded video due (hard deadline) | Thursday | April 15, 2027 |
| Workshop day | [TBD] | [TBD], May 2027 |
All deadlines are 11:59 pm UTC-12h (anywhere on earth).