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 2025 · Retrospective · Covers 5 August 2025 11:00 to 12 August 2025 11:00 UTC

Read on the web

Top developments

Google’s August Patch Fixes Two Qualcomm Vulnerabilities Exploited in the Wild

The Hacker News · Incident · Remediation

OpenHands and the Lethal Trifecta: How Prompt Injection Can Leak Access Tokens

Embrace The Red · Incident · AI Security

How Devin AI Can Leak Your Secrets via Multiple Means

Embrace The Red · Incident · AI Security

AI Kill Chain in Action: Devin AI Exposes Ports to the Internet with Prompt Injection

Embrace The Red · Research · AI Security

CISA orders fed agencies to patch new Exchange flaw by Monday

BleepingComputer · Advisory · Remediation

SAST

WhyFlow: Interrogative Debugger for Sensemaking Taint Analysis

arXiv · Research · SAST

ScamDetect: Towards a Robust, Agnostic Framework to Uncover Threats in Smart Contracts

arXiv · Research · SAST

OpenSSF at DEF CON 33: AI Cyber Challenge (AIxCC), MLSecOps, and Securing Critical Infrastructure

OpenSSF · Analysis · Remediation, SAST

Large Language Models Versus Static Code Analysis Tools: A Systematic Benchmark for Vulnerability Detection

arXiv · Research · SAST, SCA

SCA

60 malicious Ruby gems downloaded 275,000 times steal credentials

BleepingComputer · Analysis · SCA

From Beginner to Builder: Understanding OpenSSF Community and Working Groups

OpenSSF · Analysis · SCA

RubyGems, PyPI Hit by Malicious Packages Stealing Credentials, Crypto, Forcing Security Changes

The Hacker News · Research · SCA

Fake WhatsApp developer libraries hide destructive data-wiping code

BleepingComputer · Analysis · SCA

DAST

ReFuzzer: Feedback-Driven Approach to Enhance Validity of LLM-Generated Test Programs

arXiv · Research · DAST

AI Is Transforming Cybersecurity Adversarial Testing - Pentera Founder’s Vision

The Hacker News · Analysis · DAST

Agentic SDLC

I Spent $500 To Test Devin AI For Prompt Injection So That You Don't Have To

Embrace The Red · Research · Agentic SDLC, AI Security

Claude Code: Data Exfiltration with DNS (CVE-2025-55284)

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

Context Engineering for Multi-Agent LLM Code Assistants Using Elicit, NotebookLM, ChatGPT, and Claude Code

arXiv · Research · Agentic SDLC

ASTRA: Autonomous Spatial-Temporal Red-teaming for AI Software Assistants

arXiv · Research · Agentic SDLC

Remediation

Trend Micro warns of Apex One zero-day exploited in attacks

BleepingComputer · Incident · Remediation

Researchers Spot Surge in Erlang/OTP SSH RCE Exploits, 70% Target OT Firewalls

The Hacker News · Research · Remediation

Researchers Detail Windows EPM Poisoning Exploit Chain Leading to Domain Privilege Escalation

The Hacker News · Research · Remediation

DARPA’s AI Cyber Challenge reveals winning models for automated vulnerability discovery and patching

CyberScoop · Analysis · Remediation

Trust

Never compromise with vulnerabilities: a comprehensive survey on AI governance

arXiv · Research · Trust

Agentic Frameworks Summary

Elastic · Analysis · AI Security, Trust

Causal Reward Adjustment: Mitigating Reward Hacking in External Reasoning via Backdoor Correction

arXiv · Research · Models, Trust

Model Context Protocol SEP-1330: Elicitation Enum Schema Improvements and Standards Compliance

Model Context Protocol · Proposal · Trust

Models

Special-Character Adversarial Attacks on Open-Source Language Model

arXiv · Research · Models

FIDELIS: Blockchain-Enabled Protection Against Poisoning Attacks in Federated Learning

arXiv · Research · Models

BadPromptFL: A Novel Backdoor Threat to Prompt-based Federated Learning in Multimodal Models

arXiv · Research · Models

Multi-Turn Jailbreaks Are Simpler Than They Seem

arXiv · Research · Models

AI Security

Securing Agentic AI: Threat Modeling and Risk Analysis for Network Monitoring Agentic AI System

arXiv · Research · AI Security

ZombAI Exploit with OpenHands: Prompt Injection To Remote Code Execution

Embrace The Red · Analysis · AI Security

Researchers Uncover GPT-5 Jailbreak and Zero-Click AI Agent Attacks Exposing Cloud and IoT Systems

The Hacker News · Research · AI Security, Models

When a Jira Ticket Can Steal Your Secrets

Simon Willison · Analysis · AI Security

Advisories in this window

  • 27 Go advisories · 1 critical · 7 high · 15 moderate · 4 low
  • 9 npm advisories · 4 high · 3 moderate · 2 low
  • 8 Maven advisories · 3 high · 5 moderate
  • 7 PyPI advisories · 3 high · 3 moderate · 1 low
  • 5 Multi-ecosystem advisories · 5 critical
  • 5 Packagist advisories · 1 critical · 3 moderate · 1 low
  • 3 crates.io advisories · 1 high · 1 moderate · 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.