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 45 of 2025 · Retrospective · Covers 28 October 2025 11:00 to 4 November 2025 12:00 UTC

Read on the web

Top developments

SesameOp: Novel backdoor uses OpenAI Assistants API for command and control

Microsoft · Research · AI Security

PhantomRaven attack floods npm with credential-stealing packages

BleepingComputer · Analysis · SAST, SCA

Suspected Nation-State Threat Actor Uses New Airstalk Malware in a Supply Chain Attack

Palo Alto Networks · Incident · SCA

Fake Solidity VSCode extension on Open VSX backdoors developers

BleepingComputer · Analysis · SCA

Hackers use RMM tools to breach freighters and steal cargo shipments

BleepingComputer · Incident · SCA

SAST

A Comprehensive Empirical Evaluation of Agent Frameworks on Code-centric Software Engineering Tasks

arXiv · Research · SAST

Security Audit of intel ICE Driver for e810 Network Interface Card

arXiv · Research · SAST

On the Difficulty of Selecting Few-Shot Examples for Effective LLM-based Vulnerability Detection

arXiv · Research · SAST

Security Vulnerabilities in AI-Generated Code: A Large-Scale Analysis of Public GitHub Repositories

arXiv · Research · Agentic SDLC, SAST

SCA

Open VSX rotates access tokens used in supply-chain malware attack

BleepingComputer · Analysis · SCA

Learnings from recent npm supply chain compromises

Datadog · Incident · SCA

Characterizing Build Compromises Through Vulnerability Disclosure Analysis

arXiv · Research · SCA

Python Foundation Adopts SPDX for Software Bill of Materials

SPDX · Analysis · SCA

DAST

A Comprehensive Evaluation and Practice of System Penetration Testing

arXiv · Research · DAST

ImmuniWeb Continuous now enables always-on, AI-powered security testing

Help Net Security · Analysis · DAST

Agentic SDLC

OpenAI releases ‘Aardvark’ security and patching model

CyberScoop · Analysis · Agentic SDLC, AI Security, Remediation

Lost in Code Generation: Reimagining the Role of Software Models in AI-driven Software Engineering

arXiv · Research · Agentic SDLC

HAFixAgent: History-Aware Agentic Program Repair

arXiv · Research · Agentic SDLC

Claude Code Can Debug Low-level Cryptography

Simon Willison · Analysis · Agentic SDLC

Remediation

Two Windows vulnerabilities, one a 0-day, are under active exploitation

Ars Technica: Security · Analysis · Remediation

CISA and NSA share tips on securing Microsoft Exchange servers

BleepingComputer · Analysis · Remediation

Lares: LLM-driven Code Slice Semantic Search for Patch Presence Testing

arXiv · Research · Remediation

Visibility Gaps: Streamlining Patching and Vulnerability Remediation

BleepingComputer · Analysis · Remediation

Trust

Exclusive: OpenAI’s Atlas browser — and others — can be tricked by manipulated web content

CyberScoop · Research · Trust

AudAgent: Automated Auditing of Privacy Policy Compliance in AI Agents

arXiv · Research · Trust

Internal Vulnerabilities, External Threats: A Grounded Framework for Enterprise Open Source Risk Governance

arXiv · Research · SCA, Trust

AAGATE: A NIST AI RMF-Aligned Governance Platform for Agentic AI

arXiv · Research · Trust

Models

AutoAdv: Automated Adversarial Prompting for Multi-Turn Jailbreaking of Large Language Models

arXiv · Research · Models

ASTRA: An Automated Framework for Strategy Discovery, Retrieval, and Evolution for Jailbreaking LLMs

arXiv · Research · Models

ShadowLogic: Backdoors in Any Whitebox LLM

arXiv · Research · Models

Reimagining Safety Alignment with An Image

arXiv · Research · Models

AI Security

Prompt Injection as an Emerging Threat: Evaluating the Resilience of Large Language Models

arXiv · Research · AI Security

"Give a Positive Review Only": An Early Investigation Into In-Paper Prompt Injection Attacks and Defenses for AI Reviewers

arXiv · Research · AI Security

GrowthHacker: Automated Off-Policy Evaluation Optimization Using Code-Modifying LLM Agents

arXiv · Research · AI Security

DRIP: Defending Prompt Injection via Token-wise Representation Editing and Residual Instruction Fusion

arXiv · Research · AI Security

Advisories in this window

  • 245 npm advisories · 240 critical · 3 high · 2 moderate
  • 29 Maven advisories · 3 high · 26 moderate
  • 21 PyPI advisories · 8 high · 12 moderate · 1 low
  • 18 Packagist advisories · 5 high · 12 moderate · 1 low
  • 11 Go advisories · 5 high · 6 moderate
  • 4 NuGet advisories · 1 critical · 3 moderate
  • 1 crates.io 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.