Long papers
Novel research contributions, up to 8 pages long
NLP/CORE Unranked
Re-Data-COLM2026 2026
Workshop on Responsibly Enabling Data for Foundation Models
The Re-Data-COLM2026 workshop, held in conjunction with COLM 2026, focuses on responsibly enabling the use of sensitive data for training foundation models. It brings together experts in data privacy, model training, and policy to explore techniques like data transformation, synthetic data generation, and constrained training paradigms that balance utility with privacy and compliance.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short papers
Preliminary studies or negative results, up to 4 pages long
Research areas in scope
01
Topics of Interest
Data Transformation: De-identification, Anonymization, PseudonymizationSynthetic Data Generation: Controlled Regeneration, Data DiversityNovel Training Paradigms: DP, Federated Learning, Architectural SolutionsEvaluation & Auditing: Privacy attack benchmarks, Utility-Privacy tradeoffsPolicy: Compliance, New regulations on data sharing
Policies worth checking twice
- Submissions are managed via OpenReview
- Accepted papers are non-archival
- Concurrent submissions are allowed
- Please follow the COLM 2026 template
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026