MaskForge: Structure-Aware Adaptive Attacks for Jailbreaking Diffusion Large Language Models

arXiv 2606.04027•c9e518bbff91807402a6d5d356764ee30f5cdbf571ef88d65558f1e9b772cd55
activation-detectioncontextual-integritycovert-influencecredential-exfiltrationdiffusion-llmfile-type-detectionforensicsformal-verificationgnn-privacyhoneytokenshpkejailbreakmaskforgemembership-privacymemory-poisoningmerkle-logmimelensmodel-integritymodel-poisoningnotarized-receiptsparameter-attacksprivacyquery-rewritingsellostarkware

Paper metadata

arXiv ID
2606.04027
Version
Not specified by this published record
Category
Computer Science — Cryptography and Security (cs.CR)

The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.

Evidence and limitations

Source ID
arxiv_cs_cr
Record identifier
c9e518bbff91807402a6d5d356764ee30f5cdbf571ef88d65558f1e9b772cd55
Enrichment time
2026-06-04T07:23:34Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.

MaskForge: Structure-Aware Adaptive Attacks for Jailbreaking Diffusion Large Language Models · Baitaphish