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 18 April 2026 Retrospective Covers 17 April 2026 00:00 to 18 April 2026 00:00 UTC Web: https://keryn.dev/headlines/daily/2026-04-18/ TOP DEVELOPMENTS [1] Breaking Opus 4.7 with ChatGPT (Hacking Claude's Memory) Embrace The Red · Research · https://embracethered.com/blog/posts/2026/breaking-opus-4.7-with-chatgpt [2] Why Third-Party Notices Are Breaking at Scale: What the Ecosystem Needs Next OpenSSF · Research · https://openssf.org/blog/2026/04/17/why-third-party-notices-are-breaking-at-scale-what-the-ecosystem-needs-next [3] Three Microsoft Defender Zero-Days Actively Exploited; Two Still Unpatched The Hacker News · Incident · https://thehackernews.com/2026/04/three-microsoft-defender-zero-days.html SCA [4] A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecycle arXiv · Research · https://arxiv.org/abs/2604.16548 MODELS [5] Benign Fine-Tuning Breaks Safety Alignment in Audio LLMs arXiv · Research · https://arxiv.org/abs/2604.16659 [6] When Safe Concepts Become Unsafe: Multi-Concept Compositional Vulnerabilities in Text-to-Image Models arXiv · Research · https://arxiv.org/abs/2604.15967 [7] Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs arXiv · Research · https://arxiv.org/abs/2604.15780 AI SECURITY [8] Who’s Minding the Agent? A New Framework for AI Identity and Access Control CoSAI · Standard / policy · https://coalitionforsecureai.org/whos-minding-the-agent-a-new-framework-for-ai-identity-and-access-control [9] Build a More Secure, Always-On Local AI Agent with OpenClaw and NVIDIA NemoClaw NVIDIA · Analysis · https://developer.nvidia.com/blog/build-a-secure-always-on-local-ai-agent-with-nvidia-nemoclaw-and-openclaw ADVISORIES IN THIS WINDOW 49 npm advisories · 8 critical · 14 high · 25 moderate · 2 low 11 Go advisories · 5 high · 5 moderate · 1 low 5 Maven advisories · 1 critical · 3 high · 1 moderate 4 Packagist advisories · 1 critical · 1 high · 1 moderate · 1 low 3 PyPI advisories · 2 critical · 1 low 1 RubyGems 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.