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 33 of 2026 · Retrospective · Covers 4 August 2026 11:00 to 11 August 2026 11:00 UTC

Read on the web

Top developments

ChainDrop supply chain compromise: Anatomy of a self-propagating worm

Microsoft · Incident · SCA

BdThemes plugins supply-chain hack creates rogue WordPress admins

BleepingComputer · Analysis · SCA

CISA: SonicWall SMA1000 flaws now exploited by ransomware gangs

BleepingComputer · Incident · Remediation

Incident Report: unsanctioned agent behaviour during cyber testing

Simon Willison · Incident · AI Security

Anthropic’s AI used fake identities, malware in rogue attack on GitHub project

Ars Technica: Security · Incident · AI Security

SAST

SkillsMetric: Mapping the Detection Boundary of Static Analysis for Malicious Agent Skills

arXiv · Research · AI Security, SAST

RustGo: Fairly Directed Greybox Fuzzing for Enforcing Rust Memory Safety

arXiv · Research · DAST, SAST

Agent-Based Test Assertion Generation via Diverse Perspective Aggregation

arXiv · Research · SAST

Hidden Ciphers and Where to Find Them: Static Discovery and Assessment of Cryptographic Assets in Software

arXiv · Research · SAST

SCA

Massive supply-chain attack compromises 440 packages under four hours

CyberScoop · Incident · SCA

Massive ChainDrop npm supply-chain attack infects hundreds of packages

BleepingComputer · Incident · SCA

TeamPCP Linked To Redis Attacks Dating Back To 2020 And Later Supply Chain Campaign

The Hacker News · Incident · SCA

“Old Father Eternity”: Shai-Hulud Worm Strikes a 6th Time – keyv/cacheable npm Supply Chain Attack

Checkmarx · Incident · SCA

DAST

OpenAI releases ChatGPT 5.6 Cyber, but it's only for approved users

BleepingComputer · Release · DAST

Show, Don't Tell: What Evo Continuous Offensive Security Found in a Real Enterprise SaaS

Snyk · Research · DAST

Generics-Aware Fuzz Target Generation for Rust Libraries via Structured API Analysis

arXiv · Research · DAST

From Documentation to Zero-day Vulnerabilities: LLM-Driven Fuzzing of JavaScript Engines in PDF Readers

arXiv · Research · DAST

Agentic SDLC

On Understanding, Identifying, and Mitigating Vulnerabilities in Agentic Large Language Models

arXiv · Research · Agentic SDLC, AI Security, Models

13 million tool calls: auditing every AI coding agent action with Elastic Agent

Elastic · Analysis · Agentic SDLC

Activation Probes Surface Code-Security Signals that the Model's Output Misses

arXiv · Research · Agentic SDLC

Auto mode is now the default in Claude Code for Pro, Max, and Team plans

Simon Willison · Analysis · Agentic SDLC, AI Security

Remediation

CVE-2026-63077: JetBrains TeamCity Deserialization of Untrusted Data Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Remediation

More than half of AI-generated patches are broken

CyberScoop · Research · Remediation

TP-Link patches Omada ZTP flaws allowing hackers to breach networks

BleepingComputer · Advisory · Remediation

CyberForge: Verified Vulnerability Injection at Repository Level for Cybersecurity Agent Training

arXiv · Research · Remediation

Trust

Advance Zero Trust for AI: New tools and guidance to secure AI agents and DevSecOps

Microsoft · Standard / policy · AI Security, Trust

Joint Lyapunov Certificates for K-Agent Generative AI Governance: Stochastic Stability, Emergent Ensemble Risk, and Zero-Knowledge Governance Attestation

arXiv · Research · Trust

Context Is Not Authority: Structured Runtime Governance for Financial Market Agents

arXiv · Research · Trust

Who Built This Model? Tracing LLM Lineage via Spectral Fingerprints in Weight Space

arXiv · Research · Trust

Models

Measuring Semantic Abstractness of SAE Features via Nonlocality

arXiv · Research · Models

Generating Attacks for LLMs with GFlowNets

arXiv · Research · Models

Stealing Reasoning Traces from Proprietary LLM APIs

arXiv · Research · Models

Agentic Harnesses: LLM-Driven Verification Layers for Robot Autonomy

arXiv · Research · Models

AI Security

AI Recommendation Poisoning: How "Ask AI" Buttons Silently Alter LLM Memory

The Hacker News · Analysis · AI Security

A Gateway Architecture for Enterprise MCP Authentication: Unifying Heterogeneous Auth, Identity Delegation, and the User / Non-User Persona Problem

arXiv · Research · AI Security

From Prompt Injection to Web Exploitation: Revisiting Classic Vulnerabilities in LLM-Integrated Applications

arXiv · Research · AI Security

Why transparent AI agents matter more than you think

CyberScoop · Analysis · AI Security

Advisories in this window

  • 1396 npm advisories · 1326 critical · 24 high · 42 moderate · 4 low
  • 70 PyPI advisories · 35 critical · 13 high · 20 moderate · 2 low
  • 35 Packagist advisories · 3 critical · 10 high · 19 moderate · 3 low
  • 27 Go advisories · 1 critical · 8 high · 14 moderate · 4 low
  • 6 Maven advisories · 1 critical · 3 high · 2 moderate
  • 1 RubyGems advisory · 1 low

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.