PREVIEW. This is how it will look in an inbox; it was not sent to you. Each real email carries its own unsubscribe link and Keryn's postal address. KERYN HEADLINES Daily, Saturday 31 January 2026 Retrospective Covers 30 January 2026 00:00 to 31 January 2026 00:00 UTC Web: https://keryn.dev/headlines/daily/2026-01-31/ TOP DEVELOPMENTS [1] Case study: Securing AI application supply chains Microsoft · Research · https://microsoft.com/en-us/security/blog/2026/01/30/case-study-securing-ai-application-supply-chains [2] Safer by Diffusion, Broken by Context: Diffusion LLM's Safety Blessing and Its Failure Mode arXiv · Research · https://arxiv.org/abs/2602.00388 [3] "Someone Hid It": Query-Agnostic Black-Box Attacks on LLM-Based Retrieval arXiv · Research · https://arxiv.org/abs/2602.00364 SAST [4] RVDebloater: Mode-based Adaptive Firmware Debloating for Robotic Vehicles arXiv · Research · https://arxiv.org/abs/2602.00270 [5] SolAgent: A Specialized Multi-Agent Framework for Solidity Code Generation arXiv · Research · https://arxiv.org/abs/2601.23009 [6] Sifting the Noise: A Comparative Study of LLM Agents in Vulnerability False Positive Filtering arXiv · Research · https://arxiv.org/abs/2601.22952 SCA [7] MITRE ATLAS (releases) v5.3.0 MITRE · Release · https://github.com/mitre-atlas/atlas-data/releases/tag/v5.3.0 [8] Uncovering Hidden Inclusions of Vulnerable Dependencies in Real-World Java Projects arXiv · Research · https://arxiv.org/abs/2601.23020 [9] Protect your AI workloads from supply chain attacks Chainguard · Analysis · https://chainguard.dev/unchained/protect-your-ai-workloads-from-supply-chain-attacks DAST [10] Rust and Go directed fuzzing with LibAFL-DiFuzz arXiv · Research · https://arxiv.org/abs/2601.22772 AGENTIC SDLC [11] Practical Security Guidance for Sandboxing Agentic Workflows and Managing Execution Risk NVIDIA · Standard / policy · https://developer.nvidia.com/blog/practical-security-guidance-for-sandboxing-agentic-workflows-and-managing-execution-risk REMEDIATION [12] From Historical Patches to Repair Plans: Outcome-Conditioned Reasoning for Repository-Level Program Repair arXiv · Research · https://arxiv.org/abs/2601.23257 [13] Do Fine-Tuned LLMs Understand Vulnerabilities? An Investigation into the Semantic Trap arXiv · Research · https://arxiv.org/abs/2601.22655 [14] SmarterMail Fixes Critical Unauthenticated RCE Flaw with CVSS 9.3 Score The Hacker News · Analysis · https://thehackernews.com/2026/01/smartermail-fixes-critical.html TRUST [15] A Structured Approach to Safety Case Construction for AI Systems arXiv · Research · https://arxiv.org/abs/2601.22773 MODELS [16] MITRE ATLAS (releases) v5.2.0 MITRE · Release · https://github.com/mitre-atlas/atlas-data/releases/tag/v5.2.0 [17] Now You Hear Me: Audio Narrative Attacks Against Large Audio-Language Models arXiv · Research · https://arxiv.org/abs/2601.23255 [18] Character as a Latent Variable in Large Language Models: A Mechanistic Account of Emergent Misalignment and Conditional Safety Failures arXiv · Research · https://arxiv.org/abs/2601.23081 AI SECURITY [19] Verifiable Manifest Signing and Transparency Enforcement for Secure MCP-Based LLM Pipelines arXiv · Research · https://arxiv.org/abs/2601.23132 [20] Whispers of Wealth: Red-Teaming Google's Agent Payments Protocol via Prompt Injection arXiv · Research · https://arxiv.org/abs/2601.22569 ADVISORIES IN THIS WINDOW 5 PyPI advisories · 1 critical · 3 high · 1 low 3 npm advisories · 1 critical · 1 high · 1 low 1 Packagist advisory · 1 moderate 1 Maven advisory · 1 high 1 NuGet advisory · 1 moderate All advisories: https://keryn.dev/headlines/advisories/ Titles, links and short excerpts only; every item links to its source. Preview: the unsubscribe link and Keryn's postal address are added to each email when it is sent.