Recoverable Identifier
advisory
doi_compliance
recoverable_identifier
DOI in the printed bibliography is fragmented by whitespace or line breaks. A longer candidate (10.1101/2025.01.30.635558v1) was visible in the surrounding text but could not be confirmed against doi.org as printed.
Paper page Integrity report arXiv Try DOI
Evidence text
Zhihan Zhou, Robert Riley, Satria Kautsar, Weimin Wu, Rob Egan, Steven Hofmeyr, Shira Goldhaber-Gordon, Mutian Yu, Harrison Ho, Fengchen Liu, Feng Chen, Rachael Morgan-Kiss, Lizhen Shi, Han Liu, and Zhong Wang. GenomeOcean: An efficient genome foundation model trained on large-scale metagenomic assemblies.bioRxiv, 2025. doi: 10.1101/2025.01.30. 635558. URL https://www.biorxiv.org/content/10.1101/2025.01.30.635558v1. DNA language-model family used as the teacher/student backbone for the GenoTrace inspiration study. 13 Appendix A Broader Impact The mechanism is dual-use. On the defender’s side it lowers the cost of auditing unauthorized distillation, particularly for providers who cannot ship logit-level watermarks because of latency, product-fit, or post-training-pipeline constraints. On the attacker’s side it informs neutralization strategies: if a marker is known, the attacker can prompt-engineer or post-filter to remove it. We have therefore framed claims as bounded empirical findings and not as a production-ready defense, and we have flagged stronger paraphrase attacks and adaptive cleaning as open problems (App. O). The held-out evaluation does not contain personal data, and the H5 user study (Section 5.4) is conducted as informal in-lab usability testing with lab members, with no external recruitment and no identifiable personal data collected. B Reproducibility All hyperparameters, system prompts, judge rubrics, and protocol details needed to reproduce the numerical cla
Evidence payload
{
"printed_excerpt": "Zhihan Zhou, Robert Riley, Satria Kautsar, Weimin Wu, Rob Egan, Steven Hofmeyr, Shira Goldhaber-Gordon, Mutian Yu, Harrison Ho, Fengchen Liu, Feng Chen, Rachael Morgan-Kiss, Lizhen Shi, Han Liu, and Zhong Wang. GenomeOcean: An efficient gen",
"reconstructed_doi": "10.1101/2025.01.30.635558v1",
"ref_index": 47,
"resolved_title": null,
"verdict_class": "incontrovertible"
}