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 41 of 2025 · Retrospective · Covers 30 September 2025 11:00 to 7 October 2025 11:00 UTC

Read on the web

Top developments

Well, Well, Well. It’s Another Day. (Oracle E-Business Suite Pre-Auth RCE Chain - CVE-2025-61882)

watchTowr · Analysis · Remediation

CVE-2013-3918: Microsoft Windows Out-of-Bounds Write Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Agentic SDLC

CVE-2021-22555: Linux Kernel Heap Out-of-Bounds Write Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Agentic SDLC

CommetJacking attack tricks Comet browser into stealing emails

BleepingComputer · Analysis · AI Security

Critical WD My Cloud bug allows remote command injection

BleepingComputer · Advisory · Remediation

SAST

NatGVD: Natural Adversarial Example Attack towards Graph-based Vulnerability Detection

arXiv · Research · Models, SAST

Detecting and Characterizing Low and No Functionality Packages in the NPM Ecosystem

arXiv · Research · SAST

MulVuln: Enhancing Pre-trained LMs with Shared and Language-Specific Knowledge for Multilingual Vulnerability Detection

arXiv · Research · SAST

QuiLL: An LLM-Based Vulnerability Assessment Framework for the Wild

arXiv · Research · SAST

SCA

Why Software Signing (Still) Matters: Trust Boundaries in the Software Supply Chain

arXiv · Research · SCA, Trust

Security leaders at Okta and Zscaler share lessons from Salesloft Drift attacks

CyberScoop · Analysis · SCA

DMark: Order-Agnostic Watermarking for Diffusion Large Language Models

arXiv · Research · SCA

TAIBOM: Bringing Trustworthiness to AI-Enabled Systems

arXiv · Research · AI-BOM, SCA

DAST

AutoPentester: An LLM Agent-based Framework for Automated Pentesting

arXiv · Research · DAST

InsightQL: Advancing Human-Assisted Fuzzing with a Unified Code Database and Parameterized Query Interface

arXiv · Research · DAST

PentestMCP: A Toolkit for Agentic Penetration Testing

arXiv · Research · DAST

MALF: A Multi-Agent LLM Framework for Intelligent Fuzzing of Industrial Control Protocols

arXiv · Research · DAST

Agentic SDLC

REFINE: Enhancing Program Repair Agents through Context-Aware Patch Refinement

arXiv · Research · Agentic SDLC, Remediation

LLM Agents for Automated Dependency Upgrades

arXiv · Research · Agentic SDLC

RedCodeAgent: Automatic Red-teaming Agent against Diverse Code Agents

arXiv · Research · Agentic SDLC, Models

SIEVE: Towards Verifiable Certification for Code-datasets

arXiv · Research · Agentic SDLC

AI-BOM

Malice in Agentland: Down the Rabbit Hole of Backdoors in the AI Supply Chain

arXiv · Research · AI-BOM, Models

Remediation

Oracle links Clop extortion attacks to July 2025 vulnerabilities

BleepingComputer · Analysis · Remediation

Vul-R2: A Reasoning LLM for Automated Vulnerability Repair

arXiv · Research · Remediation

Redis warns of critical flaw impacting thousands of instances

BleepingComputer · Analysis · Remediation

Semantics-Aligned, Curriculum-Driven, and Reasoning-Enhanced Vulnerability Repair Framework

arXiv · Research · Remediation, SCA

Trust

Autonomy Reshapes How Personalization Affects Privacy Concerns and Trust in LLM Agents

arXiv · Research · Trust

AgentHub: A Registry for Discoverable, Verifiable, and Reproducible AI Agents

arXiv · Research · Trust

Breaking the Code: Security Assessment of AI Code Agents Through Systematic Jailbreaking Attacks

arXiv · Research · Agentic SDLC, Models, Trust

A Call to Action for a Secure-by-Design Generative AI Paradigm

arXiv · Research · AI Security, Models, Trust

Models

Toward a Safer Web: Multilingual Multi-Agent LLMs for Mitigating Adversarial Misinformation Attacks

arXiv · Research · Models

LatentBreak: Jailbreaking Large Language Models through Latent Space Feedback

arXiv · Research · Models

Correcting Prompt Dependence in LLM Benchmarks: A Bayesian Hierarchical Model with Embedding-Space Clustering

arXiv · Research · Models

AutoDAN-Reasoning: Enhancing Strategies Exploration based Jailbreak Attacks with Test-Time Scaling

arXiv · Research · Models

AI Security

Adversarial Reinforcement Learning for Large Language Model Agent Safety

arXiv · Research · AI Security

Indirect Prompt Injections: Are Firewalls All You Need, or Stronger Benchmarks?

arXiv · Research · AI Security

RL Is a Hammer and LLMs Are Nails: A Simple Reinforcement Learning Recipe for Strong Prompt Injection

arXiv · Research · AI Security, Models

Unified Threat Detection and Mitigation Framework (UTDMF): Combating Prompt Injection, Deception, and Bias in Enterprise-Scale Transformers

arXiv · Research · AI Security, Models

Advisories in this window

  • 178 npm advisories · 163 critical · 7 high · 4 moderate · 4 low
  • 12 Go advisories · 8 high · 4 moderate
  • 10 Maven advisories · 2 critical · 3 high · 5 moderate
  • 9 PyPI advisories · 1 critical · 2 high · 5 moderate · 1 low
  • 8 Packagist advisories · 2 high · 1 moderate · 5 low
  • 2 crates.io advisories · 1 critical · 1 low
  • 1 NuGet 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.