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 Weekly, week 11 of 2026 Retrospective Covers 3 March 2026 12:00 to 10 March 2026 11:00 UTC Web: https://keryn.dev/headlines/weekly/2026-W11/ TOP DEVELOPMENTS [1] Feds take notice of iOS vulnerabilities exploited under mysterious circumstances Ars Technica: Security · Incident · https://arstechnica.com/security/2026/03/cisa-adds-3-ios-flaws-to-its-catalog-of-known-exploited-vulnerabilities [2] Clinejection — Compromising Cline's Production Releases just by Prompting an Issue Triager Simon Willison · Incident · https://simonwillison.net/2026/Mar/6/clinejection [3] AI fights and more attacks on dev infrastructure: Last Week in AppSec for 4. March 2026 Checkmarx · Incident · https://checkmarx.com/zero-post/ai-fights-and-more-attacks-on-dev-infrastructure-last-week-in-appsec-for-4-march-2026 [4] Malicious npm Package Posing as OpenClaw Installer Deploys RAT, Steals macOS Credentials The Hacker News · Research · https://thehackernews.com/2026/03/malicious-npm-package-posing-as.html [5] Fake Laravel Packages on Packagist Deploy RAT on Windows, macOS, and Linux The Hacker News · Research · https://thehackernews.com/2026/03/fake-laravel-packages-on-packagist.html SAST [6] SmartGraphical: A Human-in-the-Loop Framework for Detecting Smart Contract Logical Vulnerabilities via Pattern-Driven Static Analysis and Visual Abstraction arXiv · Research · https://arxiv.org/abs/2603.08580 [7] SCAFFOLD-CEGIS: Preventing Latent Security Degradation in LLM-Driven Iterative Code Refinement arXiv · Research · https://arxiv.org/abs/2603.08520 SCA [8] SBOMs into Agentic AIBOMs: Schema Extensions, Agentic Orchestration, and Reproducibility Evaluation arXiv · Research · https://arxiv.org/abs/2603.10057 [9] Learning the APT Kill Chain: Temporal Reasoning over Provenance Data for Attack Stage Estimation arXiv · Research · https://arxiv.org/abs/2603.07560 [10] Malicious AI Assistant Extensions Harvest LLM Chat Histories Microsoft · Analysis · https://microsoft.com/en-us/security/blog/2026/03/05/malicious-ai-assistant-extensions-harvest-llm-chat-histories [11] Can the Security Platform Finally Deliver for the Mid-Market? The Hacker News · Analysis · https://thehackernews.com/2026/03/can-security-platform-finally-deliver.html DAST [12] Coverage-Guided Multi-Agent Harness Generation for Java Library Fuzzing arXiv · Research · https://arxiv.org/abs/2603.08616 [13] SemFuzz: A Semantics-Aware Fuzzing Framework for Network Protocol Implementations arXiv · Research · https://arxiv.org/abs/2603.05989 AGENTIC SDLC [14] Arbiter: Detecting Interference in LLM Agent System Prompts arXiv · Research · https://arxiv.org/abs/2603.08993 [15] Secure agentic AI for your Frontier Transformation Microsoft · Analysis · https://microsoft.com/en-us/security/blog/2026/03/09/secure-agentic-ai-for-your-frontier-transformation [16] Turn: A Language for Agentic Computation arXiv · Research · https://arxiv.org/abs/2603.08755 [17] ESAA-Security: An Event-Sourced, Verifiable Architecture for Agent-Assisted Security Audits of AI-Generated Code arXiv · Research · https://arxiv.org/abs/2603.06365 REMEDIATION [18] Why LLMs Fail: A Failure Analysis and Partial Success Measurement for Automated Security Patch Generation arXiv · Research · https://arxiv.org/abs/2603.10072 [19] OSS-CRS: Liberating AIxCC Cyber Reasoning Systems for Real-World Open-Source Security arXiv · Research · https://arxiv.org/abs/2603.08566 [20] Patch Validation in Automated Vulnerability Repair arXiv · Research · https://arxiv.org/abs/2603.06858 [21] Patch diff to SYSTEM Elastic · Analysis · https://elastic.co/security-labs/blog/patch-diff-to-system TRUST [22] Governance Architecture for Autonomous Agent Systems: Threats, Framework, and Engineering Practice arXiv · Research · https://arxiv.org/abs/2603.07191 [23] Proof-of-Guardrail in AI Agents and What (Not) to Trust from It arXiv · Research · https://arxiv.org/abs/2603.05786 [24] Steering Frozen LLMs: Adaptive Social Alignment via Online Prompt Routing arXiv · Research · https://arxiv.org/abs/2603.15647 [25] The Controllability Trap: A Governance Framework for Military AI Agents arXiv · Research · https://arxiv.org/abs/2603.03515 MODELS [26] Reasoning-Oriented Programming: Chaining Semantic Gadgets to Jailbreak Large Vision Language Models arXiv · Research · https://arxiv.org/abs/2603.09246 [27] ADVERSA: Measuring Multi-Turn Guardrail Degradation and Judge Reliability in Large Language Models arXiv · Research · https://arxiv.org/abs/2603.10068 [28] The Struggle Between Continuation and Refusal: A Mechanistic Analysis of the Continuation-Triggered Jailbreak in LLMs arXiv · Research · https://arxiv.org/abs/2603.08234 [29] DistillGuard: Evaluating Defenses Against LLM Knowledge Distillation arXiv · Research · https://arxiv.org/abs/2603.07835 AI SECURITY [30] Towards Modeling Cybersecurity Behavior of Humans in Organizations arXiv · Research · https://arxiv.org/abs/2603.08484 [31] SlowBA: An efficiency backdoor attack towards VLM-based GUI agents arXiv · Research · https://arxiv.org/abs/2603.08316 [32] VoiceSHIELD-Small: Real-Time Malicious Speech Detection and Transcription arXiv · Research · https://arxiv.org/abs/2603.07708 [33] From Thinker to Society: Security in Hierarchical Autonomy Evolution of AI Agents arXiv · Research · https://arxiv.org/abs/2603.07496 ADVISORIES IN THIS WINDOW 240 npm advisories · 64 critical · 65 high · 98 moderate · 13 low 64 Go advisories · 14 critical · 27 high · 22 moderate · 1 low 34 PyPI advisories · 3 critical · 20 high · 10 moderate · 1 low 32 Packagist advisories · 4 critical · 7 high · 19 moderate · 2 low 22 crates.io advisories · 8 critical · 8 high · 4 moderate · 2 low 16 Maven advisories · 5 critical · 7 high · 3 moderate · 1 low 1 RubyGems advisory · 1 moderate 1 Multi-ecosystem advisory · 1 high 1 GitHub Actions 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.