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, Thursday 5 March 2026 Retrospective Covers 4 March 2026 00:00 to 5 March 2026 00:00 UTC Web: https://keryn.dev/headlines/daily/2026-03-05/ TOP DEVELOPMENTS [1] 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 [2] Mail2Shell zero-click attack lets hackers hijack FreeScout mail servers BleepingComputer · Incident · https://bleepingcomputer.com/news/security/mail2shell-zero-click-attack-lets-hackers-hijack-freescout-mail-servers [3] US and EU police shut down LeakBase, a site accused of sharing stolen passwords and hacking tools TechCrunch: Security · Incident · https://techcrunch.com/2026/03/04/u-s-and-eu-police-shut-down-leakbase-a-site-accused-of-sharing-stolen-passwords-and-hacking-tools AGENTIC SDLC [4] Internal Safety Collapse in LLMs and Coding Agents arXiv · Research · https://arxiv.org/abs/2603.23509 REMEDIATION [5] Lock and unlock draft repository security advisories GitHub · Analysis · https://github.blog/changelog/2026-03-04-lock-and-unlock-draft-repository-security-advisories [6] Cisco warns of max severity Secure FMC flaws giving root access BleepingComputer · Analysis · https://bleepingcomputer.com/news/security/cisco-warns-of-max-severity-secure-fmc-flaws-giving-root-access [7] Code Fingerprints: Disentangled Attribution of LLM-Generated Code arXiv · Research · https://arxiv.org/abs/2603.04212 TRUST [8] Steering Frozen LLMs: Adaptive Social Alignment via Online Prompt Routing arXiv · Research · https://arxiv.org/abs/2603.15647 [9] New RFP Template for AI Usage Control and AI Governance The Hacker News · Analysis · https://thehackernews.com/2026/03/new-rfp-template-for-ai-usage-control.html MODELS [10] Efficient Refusal Ablation in LLM through Optimal Transport arXiv · Research · https://arxiv.org/abs/2603.04355 [11] Tuning Just Enough: Lightweight Backdoor Attacks on Multi-Encoder Diffusion Models arXiv · Research · https://arxiv.org/abs/2603.04064 [12] When Safety Becomes a Vulnerability: Exploiting LLM Alignment Homogeneity for Transferable Blocking in RAG arXiv · Research · https://arxiv.org/abs/2603.03919 AI SECURITY [13] The MCP AuthN/Z Nightmare Doyensec · Analysis · https://blog.doyensec.com/2026/03/05/mcp-nightmare.html [14] MIND: Unified Inquiry and Diagnosis RL with Criteria Grounded Clinical Supports for Psychiatric Consultation arXiv · Research · https://arxiv.org/abs/2603.03677 [15] Image-based Prompt Injection: Hijacking Multimodal LLMs through Visually Embedded Adversarial Instructions arXiv · Research · https://arxiv.org/abs/2603.03637 ADVISORIES IN THIS WINDOW 23 npm advisories · 3 critical · 8 high · 10 moderate · 2 low 10 Go advisories · 3 critical · 5 high · 2 moderate 8 PyPI advisories · 7 high · 1 moderate 8 crates.io advisories · 2 critical · 3 high · 3 moderate 8 Packagist advisories · 2 high · 5 moderate · 1 low 4 Maven advisories · 1 critical · 2 high · 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.