PuzzleMark: Implicit Jigsaw Learning for Robust Code Dataset Watermarking in Neural Code Completion Models DOI: 10.1145/3797122 Journal: PACMSE Volume: 3 FSE FSE027 7 fse26maina-p509-p CCS: Security and privacy Digital rights management CCS: Software and its engineering Software libraries and repositories CCS: Computing methodologies Artificial intelligence
Haocheng Huang, Yuchen Chen, Weisong Sun, Peizhuo Lv, Yuan Xiao, Chunrong Fang, Yang Liu, Xiaofang Zhang · 2026
Read PuzzleMark: Implicit Jigsaw Learning for Robust Code Dataset Watermarking in Neural Code Completion Models DOI: 10.1145/3797122 Journal: PACMSE Volume: 3 FSE FSE027 7 fse26maina-p509-p CCS: Security and privacy Digital rights management CCS: Software and its engineering Software libraries and repositories CCS: Computing methodologies Artificial intelligence by Haocheng Huang, Yuchen Chen, Weisong Sun, Peizhuo Lv, Yuan Xiao, Chunrong Fang, Yang Liu, Xiaofang Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-27677.txt.
Description
Huang and seven colleagues offer PuzzleMark, a tough watermarking method for code data sets. To reduce leak risk, it picks carrier snippets by code complexity, choosing highly fit snippets for marking. It replaces the old co-occurrence pattern with a novel joining pattern through variable name joining, placing watermarks to fit the code while keeping design simple. Checks use Fisher's exact test under black-box settings, reaching 100 percent success with zero false alarms and tiny performance impact.
Catalog record
| CREATOR | Haocheng Huang, Yuchen Chen, Weisong Sun, Peizhuo Lv, Yuan Xiao, Chunrong Fang, Yang Liu, Xiaofang Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 75252 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.27677