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 17 of 2026 · Retrospective · Covers 14 April 2026 11:00 to 21 April 2026 11:00 UTC

Read on the web

Top developments

Microsoft Issues Patches for SharePoint Zero-Day and 168 Other New Vulnerabilities

The Hacker News · Advisory · Remediation

Vuln in Google’s Antigravity AI agent manager could escape sandbox, give attackers remote code execution

CyberScoop · Analysis · AI Security, Remediation

Context.ai OAuth Token Compromise

Wiz · Incident · SCA

Cisco says critical Webex Services flaw requires customer action

BleepingComputer · Analysis · Remediation

Breaking Opus 4.7 with ChatGPT (Hacking Claude's Memory)

Embrace The Red · Research · AI Security, Models

SAST

SAGE: Signal-Amplified Guided Embeddings for LLM-based Vulnerability Detection

arXiv · Research · SAST

TitanCA: Lessons from Orchestrating LLM Agents to Discover 100+ CVEs

arXiv · Research · SAST

SDLLMFuzz: Dynamic-static LLM-assisted greybox fuzzing for structured input programs

arXiv · Research · DAST, SAST

NFTDELTA: Detecting Permission Control Vulnerabilities in NFT Contracts through Multi-View Learning

arXiv · Research · SAST

SCA

Why Third-Party Notices Are Breaking at Scale: What the Ecosystem Needs Next

OpenSSF · Research · SCA

Secure Your Spot: The OpenSSF Community Day North America 2026 Agenda is Live!

OpenSSF · Analysis · SCA

Weaponizing the Commons: A Taxonomy and Detection Framework of Abuse on GitHub

arXiv · Research · SCA

Original Sin of npm: A Study on Vulnerability Propagation in JavaScript Dependency Networks

arXiv · Research · SCA

DAST

Towards Optimal Agentic Architectures for Offensive Security Tasks

arXiv · Research · AI Security, DAST

Capturing Monetarily Exploitable Vulnerability in Smart Contracts via Auditor Knowledge-Learning Fuzzing

arXiv · Research · DAST

MASFuzzer: Fuzz Driver Generation and Adaptive Scheduling via Multidimensional API Sequences

arXiv · Research · DAST

CFITSIO Fuzzing: Memory Corruptions and a Codex-Assisted Pipeline

Doyensec · Analysis · Agentic SDLC, DAST

Agentic SDLC

Towards an Agentic LLM-based Approach to Requirement Formalization from Unstructured Specifications

arXiv · Research · Agentic SDLC

Owner-Harm: A Missing Threat Model for AI Agent Safety

arXiv · Research · Agentic SDLC, AI Security, Models

Project Prometheus: Bridging the Intent Gap in Agentic Program Repair via Reverse-Engineered Executable Specifications

arXiv · Research · Agentic SDLC

KnowPilot: Your Knowledge-Driven Copilot for Domain Tasks

arXiv · Research · Agentic SDLC

Remediation

Microsoft releases emergency updates to fix Windows Server issues

BleepingComputer · Analysis · Remediation

RAVEN: Retrieval-Augmented Vulnerability Exploration Network for Memory Corruption Analysis in User Code and Binary Programs

arXiv · Research · Remediation

Terminal Wrench: A Dataset of 331 Reward-Hackable Environments and 3,632 Exploit Trajectories

arXiv · Research · Remediation

SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair

arXiv · Research · Remediation

Trust

The Open-Weight Paradox: Why Restricting Access to AI Models May Undermine the Safety It Seeks to Protect

arXiv · Research · Trust

Beyond Task Success: An Evidence-Synthesis Framework for Evaluating, Governing, and Orchestrating Agentic AI

arXiv · Research · Trust

Governed MCP: Kernel-Level Tool Governance for AI Agents via Logit-Based Safety Primitives

arXiv · Research · Trust

CapSeal: Capability-Sealed Secret Mediation for Secure Agent Execution

arXiv · Research · AI Security, Trust

Models

ProjLens: Unveiling the Role of Projectors in Multimodal Model Safety

arXiv · Research · Models

Mechanistic Anomaly Detection via Functional Attribution

arXiv · Research · Models

An Empirical Study of Multi-Generation Sampling for Jailbreak Detection in Large Language Models

arXiv · Research · Models

Towards Understanding the Robustness of Sparse Autoencoders

arXiv · Research · Models

AI Security

Who’s Minding the Agent? A New Framework for AI Identity and Access Control

CoSAI · Standard / policy · AI Security

Temporal UI State Inconsistency in Desktop GUI Agents: Formalizing and Defending Against TOCTOU Attacks on Computer-Use Agents

arXiv · Research · AI Security

Beyond Pattern Matching: Seven Cross-Domain Techniques for Prompt Injection Detection

arXiv · Research · AI Security

SafeAgent: A Runtime Protection Architecture for Agentic Systems

arXiv · Research · AI Security

Advisories in this window

  • 211 npm advisories · 111 critical · 43 high · 53 moderate · 4 low
  • 61 Go advisories · 9 critical · 28 high · 18 moderate · 6 low
  • 59 PyPI advisories · 5 critical · 11 high · 32 moderate · 11 low
  • 52 Packagist advisories · 5 critical · 21 high · 24 moderate · 2 low
  • 25 NuGet advisories · 2 critical · 4 high · 10 moderate · 9 low
  • 22 Maven advisories · 5 critical · 7 high · 10 moderate
  • 9 crates.io advisories · 2 critical · 2 high · 2 moderate · 3 low
  • 5 RubyGems advisories · 3 high · 2 moderate
  • 1 Multi-ecosystem advisory · 1 moderate

All advisories

Titles, links and short excerpts only; every item links to its source. Selection is automatic; the editor's note is AI-drafted and checked as described. Preview: the unsubscribe link and Keryn's postal address are added to each email when it is sent.