Logit Laundering: Evading Data-Use Auditing in Large Language Models

Summary

Researchers propose Logit Laundering, an evasion framework in which a non-compliant LLM provider manipulates model outputs at inference time to suppress watermark-induced statistical signals used by data-use auditors. The method reshapes logits with a plug-and-play statistical bias estimator and a confidence-aware scaling module, leaving model parameters unchanged. Extensive evaluations reportedly evade various state-of-the-art auditing methods while preserving utility on downstream tasks.

Related items

• MediumCVE-2026-124588: huggingface finetrainers code injection via Library argumentSimilar attack · NVD/CVE Database
• MediumTaiwan-Based Hacker Used BOSSAI AI Agents to Steal Taipei Emergency Dispatch RecordsSimilar attack · The Hacker News
• MediumOn the hunt for ARTEX: understanding agentic attacks in AWS environmentsSimilar attack · Datadog Security Labs
• CriticalCVE-2026-102895: GPUStack Improper Authentication VulnerabilitySimilar attack · CISA Known Exploited Vulnerabilities
• HighCVE-2026-97677: IBM Langflow OSS arbitrary file write and read by authenticated flow authorsSimilar attack · NVD/CVE Database