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 35 of 2025 · Retrospective · Covers 19 August 2025 11:00 to 26 August 2025 11:00 UTC

Read on the web

Top developments

Apple fixes new zero-day flaw exploited in targeted attacks

BleepingComputer · Incident · Agentic SDLC, Remediation

Hijacking Windsurf: How Prompt Injection Leaks Developer Secrets

Embrace The Red · Incident · Agentic SDLC, AI Security

MCP vulnerability case study: SQL injection in the Postgres MCP server

Datadog · Research · AI Security

Microsoft releases emergency updates to fix Windows recovery

BleepingComputer · Research · Remediation

Apache ActiveMQ Flaw Exploited to Deploy DripDropper Malware on Cloud Linux Systems

The Hacker News · Incident · Remediation

SAST

MoveScanner: Analysis of Security Risks of Move Smart Contracts

arXiv · Research · SAST

Trust Me, I Know This Function: Hijacking LLM Static Analysis using Bias

arXiv · Research · SAST

When Machine Learning Meets Vulnerability Discovery: Challenges and Lessons Learned

arXiv · Research · SAST

Assessing the Quality and Security of AI-Generated Code: A Quantitative Analysis

arXiv · Research · SAST

SCA

Malicious Go Module Poses as SSH Brute-Force Tool, Steals Credentials via Telegram Bot

The Hacker News · Research · SCA

What Not to Miss at Open Source Summit & OpenSSF Community Day Europe

OpenSSF · Analysis · SCA

NodeShield: Runtime Enforcement of Security-Enhanced SBOMs for Node.js

arXiv · Research · SAST, SCA

DAST

Automation Is Redefining Pentest Delivery

The Hacker News · Analysis · DAST

Agentic SDLC

Safeguarding VS Code against prompt injections

GitHub · Analysis · Agentic SDLC, AI Security

A.S.E: A Repository-Level Benchmark for Evaluating Security in AI-Generated Code

arXiv · Research · Agentic SDLC

Sneaking Invisible Instructions by Developers in Windsurf

Embrace The Red · Analysis · Agentic SDLC, AI Security

CelloAI: Leveraging Large Language Models for HPC Software Development in High Energy Physics

arXiv · Research · Agentic SDLC

Remediation

Microsoft: August Windows updates cause severe streaming issues

BleepingComputer · Analysis · Remediation

Microsoft asks customers for feedback on reported SSD failures

BleepingComputer · Analysis · Remediation

Amazon Q Developer: Remote Code Execution with Prompt Injection

Embrace The Red · Analysis · Agentic SDLC, AI Security, Remediation

Microsoft shares workaround for Teams "couldn't connect" error

BleepingComputer · Analysis · Remediation

Trust

A Survey of Threats Against Voice Authentication and Anti-Spoofing Systems

arXiv · Research · Models, Trust

BadFU: Backdoor Federated Learning through Adversarial Machine Unlearning

arXiv · Research · Models, Trust

Models

Stand on The Shoulders of Giants: Building JailExpert from Previous Attack Experience

arXiv · Research · Models

Speculative Safety-Aware Decoding

arXiv · Research · Models

Tricking LLM-Based NPCs into Spilling Secrets

arXiv · Research · AI Security, Models

Attacking LLMs and AI Agents: Advertisement Embedding Attacks Against Large Language Models

arXiv · Research · AI Security, Models

AI Security

Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet

Simon Willison · Analysis · AI Security

PhantomLint: Principled Detection of Hidden LLM Prompts in Structured Documents

arXiv · Research · AI Security

Mind the Gap: Time-of-Check to Time-of-Use Vulnerabilities in LLM-Enabled Agents

arXiv · Research · AI Security

Towards Log Analysis with AI Agents: Cowrie Case Study

arXiv · Research · AI Security

Advisories in this window

  • 41 Maven advisories · 1 critical · 4 high · 32 moderate · 4 low
  • 16 npm advisories · 5 critical · 3 high · 7 moderate · 1 low
  • 15 Go advisories · 1 critical · 1 high · 10 moderate · 3 low
  • 14 Packagist advisories · 4 high · 9 moderate · 1 low
  • 13 PyPI advisories · 6 high · 7 moderate
  • 7 NuGet advisories · 3 high · 1 moderate · 3 low
  • 1 Multi-ecosystem advisory · 1 moderate
  • 1 crates.io advisory · 1 low
  • 1 RubyGems advisory · 1 critical

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.