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, Wednesday 21 October 2026 · Retrospective · Covers 20 October 2026 00:00 to 21 October 2026 00:00 UTC

Read on the web

Top developments

From Phishing to LLMjacking: Threat Actors Targeting Cloud Billing Accounts for LLM Usage

Wiz · Incident · AI Security

8 eMudhra certificates are compromised. Here's what you need to know

Cloudflare: Security · Incident · Remediation

Qilin and INC Ransomware Groups Exploit Palo Alto Networks Flaw to Bypass MFA

The Hacker News · Incident · Remediation

Hackers Use Autonomous AI Agents to Breach Company Websites, Target Over 1,200 Sites

The Hacker News · Incident · AI Security

Critical FortiSIEM Flaw Gets Public Exploit PoC for Unauthenticated Remote Code Execution

The Hacker News · Research · Remediation

SAST

How Much Does the Harness Matter, and for Whom? A Multi-Model Study of Context in LLM-Driven Vulnerability Discovery

arXiv · Research · Agentic SDLC, SAST

Self-Evolving Static Analysis via LLM Pseudo-Labeling

arXiv · Research · SAST

Raise or Check? Predicting the Necessity of Additional Fuzzing for Continuous Fuzzing

arXiv · Research · DAST, SAST

SCA

Sign Here: Attributing a Source for a More Trustworthy Software Supply Chain

arXiv · Research · SCA

Dependabot alerts REST API now includes dependency versions

GitHub · Analysis · SCA

GitHub package firewall in public preview

GitHub · Analysis · SCA

DAST

Introducing pathrunner: An AWS privilege escalation framework

Datadog · Standard / policy · DAST

Agentic SDLC

A Concurrent Structured Record Mirrored the Prose It Accompanied: A Pre-Registered Conceptual Replication and Extension of the Hide Pending Tool Call Scenario

arXiv · Research · Agentic SDLC

Agentic Verified Optimization of Real-World Software

arXiv · Research · Agentic SDLC

Act on Your Behalf, Beyond Your Authorization: Benchmarking LLM Agent Authorization Alignment

arXiv · Research · Agentic SDLC, Trust

Remediation

php2shell: Assessing Elastic detection coverage relevant to CVE-2026-109618 and CVE-2026-109619

Elastic · Analysis · Remediation

PHP warns admins of two critical flaws, emergency patches coming tomorrow

BleepingComputer · Analysis · Remediation

There's only so much code: the rise in vulnerability numbers

Cisco · Analysis · Remediation

Trust

Reusing Assurance Evidence in AI Pipelines: Validity Conditions, a Fail-Closed Release Gate, and a Replay of 714 Public Data Version Control Pipelines

arXiv · Research · Trust

KNOX: Hardware-assisted Regulatory Sandbox for Generative Models

arXiv · Research · Trust

The 9 best KnowBe4 alternatives in 2026

Aikido Security · Analysis · Trust

Models

Behavior Beyond Text: Evaluating Refusal Transfer Failure in LLM Agents

arXiv · Research · Models

Black-Box Red-Teaming of Multi-Agent Systems via Reinforcement Learning

arXiv · Research · AI Security, Models

AI Security Literacy: Ecosystem-Based Competency Framework and K-12 Learning Platform

arXiv · Research · Models

AI Security

Combining Long-Term Memory and LLMs Comes at a High Cost: From Prompt Injection to Persistent Compromise

arXiv · Research · AI Security, Models

After the Attacker Leaves: Benchmarking Carryover Attacks on LLM Agents

arXiv · Research · AI Security

Context-Aware Prompt Injection Detection via Small Language Models in RAG Pipelines: nokast-secureRAG

arXiv · Research · AI Security

Advisories in this window

  • 79 npm advisories · 45 critical · 11 high · 19 moderate · 4 low
  • 28 PyPI advisories · 15 critical · 7 high · 5 moderate · 1 low
  • 5 Maven advisories · 2 high · 3 moderate
  • 1 Packagist advisory · 1 moderate
  • 1 Go advisory · 1 high
  • 1 RubyGems advisory · 1 high

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.