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

Monthly, July 2025 · Retrospective · Covers July 2025

Read on the web

Top developments

CVE-2025-53770: Microsoft SharePoint Deserialization of Untrusted Data Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Remediation

Google fixes actively exploited sandbox escape zero day in Chrome

BleepingComputer · Incident · Remediation

CISA tags Citrix Bleed 2 as exploited, gives agencies a day to patch

BleepingComputer · Incident · Remediation

Pre-Auth SQL Injection to RCE - Fortinet FortiWeb Fabric Connector (CVE-2025-25257)

watchTowr · Analysis · Remediation

Apple patches security flaw exploited in Chrome zero-day attacks

BleepingComputer · Advisory · Remediation

Microsoft: SharePoint flaws exploited in Warlock ransomware attacks

BleepingComputer · Incident · Remediation

Cisco: Maximum-severity ISE RCE flaws now exploited in attacks

BleepingComputer · Incident · Remediation

Microsoft Patch Tuesday addresses 130 vulnerabilities, none actively exploited

CyberScoop · Incident · Remediation

CVE-2025-49704: Microsoft SharePoint Code Injection Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Remediation

Wiz Research Uncovers Critical Vulnerability in AI Vibe Coding platform Base44 Allowing Unauthorized Access to Private Applications

Wiz · Research · Agentic SDLC, Remediation

SAST

Modeling CORS frameworks with CodeQL to find security vulnerabilities

GitHub · Research · SAST

SAEL: Leveraging Large Language Models with Adaptive Mixture-of-Experts for Smart Contract Vulnerability Detection

arXiv · Research · SAST

Program Analysis for High-Value Smart Contract Vulnerabilities: Techniques and Insights

arXiv · Research · SAST, Trust

Information Security Based on LLM Approaches: A Review

arXiv · Research · SAST

When LLMs Copy to Think: Uncovering Copy-Guided Attacks in Reasoning LLMs

arXiv · Research · SAST

SCA

PyPI Users Email Phishing Attack

Python Package Index · Analysis · SCA

NVIDIAScape - Critical NVIDIA AI Vulnerability: A Three-Line Container Escape in NVIDIA Container Toolkit (CVE-2025-23266)

Wiz · Analysis · SCA

Introducing OSS Rebuild: Open Source, Rebuilt to Last

Google · Announcement · SCA, Trust

Supply-chain attacks on open source software are getting out of hand

Ars Technica: Security · Incident · SCA

Hackers breach Toptal GitHub account, publish malicious npm packages

BleepingComputer · Incident · SCA

DAST

Repeater Strike: manual testing, amplified

PortSwigger Research · Research · DAST

Search-Based Fuzzing For RESTful APIs That Use MongoDB

arXiv · Research · DAST

Mut4All: Fuzzing Compilers via LLM-Synthesized Mutators Learned from Bug Reports

arXiv · Research · DAST

An Enumerative Embedding of the Python Type System in ACL2s

arXiv · Research · DAST

Scheduzz: Constraint-based Fuzz Driver Generation with Dual Scheduling

arXiv · Research · DAST

Agentic SDLC

Flaw in Gemini CLI AI coding assistant allowed stealthy code execution

BleepingComputer · Analysis · Agentic SDLC, AI Security

Amazon AI coding agent hacked to inject data wiping commands

BleepingComputer · Incident · Agentic SDLC

The zero-day that could've compromised every Cursor and Windsurf user

BleepingComputer · Incident · Agentic SDLC, AI Security

The Month of AI Bugs 2025

Embrace The Red · Analysis · Agentic SDLC, Remediation

Multi-Stage Prompt Inference Attacks on Enterprise LLM Systems

arXiv · Research · Agentic SDLC, AI Security

Remediation

SonicWall urges admins to patch critical RCE flaw in SMA 100 devices

BleepingComputer · Advisory · Remediation

Hackers exploit SAP NetWeaver bug to deploy Linux Auto-Color malware

BleepingComputer · Analysis · Remediation

Sploitlight: Analyzing a Spotlight-based macOS TCC vulnerability

Microsoft · Analysis · Remediation

Mitel warns of critical MiVoice MX-ONE authentication bypass flaw

BleepingComputer · Analysis · Remediation

US nuclear weapons agency hacked in Microsoft SharePoint attacks

BleepingComputer · Incident · Remediation

Trust

Case Study: Google Secures Machine Learning Models with sigstore

OpenSSF · Research · Models, Trust

Model-Based Soft Maximization of Suitable Metrics of Long-Term Human Power

arXiv · Research · Trust

Building to Last: A New Governance Model for MCP

Model Context Protocol · Analysis · Trust

CoSAI x AI Security Forum: Co-Hosted Reception & Dinner During Black Hat, DEF CON

CoSAI · Analysis · Trust

An Overview of the Risk-based Model of AI Governance

arXiv · Research · Trust

Models

Measuring Harmfulness of Computer-Using Agents

arXiv · Research · Models

Exploiting Synergistic Cognitive Biases to Bypass Safety in LLMs

arXiv · Research · Models

Universally Unfiltered and Unseen:Input-Agnostic Multimodal Jailbreaks against Text-to-Image Model Safeguards

arXiv · Research · Models

Strategic Deflection: Defending LLMs from Logit Manipulation

arXiv · Research · Models

Secure Tug-of-War (SecTOW): Iterative Defense-Attack Training with Reinforcement Learning for Multimodal Model Security

arXiv · Research · Models

AI Security

Security Challenges in AI Agent Deployment: Insights from a Large Scale Public Competition

arXiv · Research · AI Security

Model Context Protocol SEP-985: Align OAuth 2.0 Protected Resource Metadata with RFC 9728

Model Context Protocol · Proposal · AI Security, SCA

Role-Aware Language Models for Secure and Contextualized Access Control in Organizations

arXiv · Research · AI Security

Counterfactual Evaluation for Blind Attack Detection in LLM-based Evaluation Systems

arXiv · Research · AI Security

Invisible Injections: Exploiting Vision-Language Models Through Steganographic Prompt Embedding

arXiv · Research · AI Security

Advisories in this window

  • 869 npm advisories · 828 critical · 22 high · 15 moderate · 4 low
  • 65 Maven advisories · 3 critical · 8 high · 50 moderate · 4 low
  • 51 PyPI advisories · 4 critical · 16 high · 26 moderate · 5 low
  • 37 Go advisories · 1 critical · 20 high · 12 moderate · 4 low
  • 22 Packagist advisories · 5 critical · 9 high · 6 moderate · 2 low
  • 8 crates.io advisories · 2 high · 3 moderate · 3 low
  • 5 RubyGems advisories · 1 critical · 4 moderate
  • 4 NuGet advisories · 2 high · 2 moderate
  • 3 Multi-ecosystem advisories · 2 high · 1 moderate
  • 3 GitHub Actions advisories · 1 critical · 2 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.