Diffusion-based image generation models produce high-quality images but may generate harmful or policy-violating content, motivating machine unlearning (MU) to selectively remove target concepts while preserving general capabilities. However, existing MU approaches often struggle with the distributed and entangled natu...
Yiran Xiahou, Siliang Qin, Kai Chen et al.· 0 citations
Agentic AI systems using the Model Context Protocol (MCP) are vulnerable to indirect prompt injection through untrusted tool outputs. We demonstrate cross-modal trust laundering, in which instructions concealed in an email-signature image using least-significant-bit (LSB), discrete cosine transform (DCT), or metadata s...
Eric Yocam, Varghese Vaidyan, Paris Kalathas et al.· 0 citations
This exploratory setting rather than population-wide detection accuracy is described, with three reviewer-package pairs and no genuine-file controls, and the findings describe this exploratory setting rather than population-wide detection accuracy.
Si-Miao Ren, Xing-Yu Shen, Tommy Duong et al.· 0 citations
Data voids are established as a practical inference-time attack surface for manipulating web-searching LLMs through strategically placed web content and highlight the urgent need for defenses against such exploits.
Michele Guerra, Wuyeh Jobe, Sanjana Nambiar et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work introduces $\mathrm{ASR}_{\mathrm{STEP}}$, which penalizes additional victim-model calls while retaining binary success judgments, and considers GCG, PAIR with a step budget, and a single-step PAP baseline on JailBreakBench across three model families.
It is demonstrated that E2E recovery alone may obscure important distinctions between encoding and decoding, motivating their separate evaluation to better understand and mitigate the safety issues caused by LLM steganography.
Om Tailor, Yu-Cheng Pan, Solaleh Mohammadi et al.· 0 citations
Modern LLM agents can retrieve information from accessible documents, and then disclose a sensitive conclusion assembled from them. This mosaic effect creates a critical risk for agents operating over confidential data. We quantify its realization with 30 synthetic cybersecurity, medical, and military examples, yieldin...
Daniel Ruiz, Luca Luceri, Emilio Ferrara· 0 citations
A systematic survey and human-coded evidence map of 118 full-text-confirmed RAG defense papers, organized by measured endpoint rather than claimed goal or defense mechanism, and the taxonomy, dataset, coding protocols, and testbed are released to enable comparable, claim-aligned RAG security evaluation.
Peiyang Liu, Xi Wang, Di Liang et al.· 0 citations
This work asks how consulting memory during dialogue changes what users go on to say, and thus what the memory comes to reveal, and across four support models, memory raises response quality but also empirical leakage, the share of attribute values correctly inferred from the record, by a margin that widens over sessio...
Giskard is presented, a scalable protocol for private and Byzantine-robust decentralized aggregation that reduces per-party communication complexity by close to three orders of magnitude over the closest competitor while exhibiting comparable robustness under up to $n/4 Byzantine parties.
Ousmane Touat, César Sabater, Mohamed Maouche et al.· 0 citations
MarkTune is introduced, a theoretically grounded on-policy fine-tuning framework that uses the GaussMark detection statistic as a reward while preserving generation quality through cross-entropy on high-quality human text or KL divergence to a stronger oracle model.
Yi-Zhou Zhao, Zhiwei Steven Wu, Adam Block· 0 citations
This work shows that, when there is a quadratic gap in computational resources between attacker and defender, every learnable classification task admits either a transferable attack or an adversarial defense, and proves a stronger result that implies this failure.
Greg Gluch, S. Goldwasser, Anastasiia Struss· 1 citation
Research Manager Mike Walker reflects on a career spanning hacker culture, cybersecurity, and AI, sharing lessons about trust, teamwork, leadership, and why solving difficult problems requires curiosity, judgment, agility—and an understanding of the rules. The post Learn the rules: The hacker mindset appeared first on Microsoft Research.
Critical flaws, compromised release channels and agentic AI threats landed within days of each other in mid-October 2026. Why “Black October” may say more about how well we now find weaknesses than about the attacks themselves. The post Black October? The Week AI Changed the Cybersecurity Conversation appeared first on GPT-Lab.
