SANS StormCast (podcast), first reported Wed 28 Oct, 03:30. Importance 4 of 5
Microsoft released out-of-band Windows 10 and 11 updates to distrust the compromised eMudhra intermediate CA certificates, and Apple now blocks them too. Command-line tools and Linux CA bundles may still trust them until updated. The episode also reports confirmed exploitation of the PHP vulnerabilities against Palo Alto devices, a token leak in Microsoft Authenticator for Android, and TLS validation flaws in the GoAhead web server.
arxiv.org, first reported Thu 29 Oct, 15:30. Importance 3 of 5
A paper by Hanzhi Liu et al. formalizes a threat model for malicious LLM API routers, with payload injection and secret exfiltration as the two attack classes. Of 28 paid and 400 free routers tested, 1 paid and 8 free routers were actively injecting malicious code. Relevant if you send API traffic through third-party routers.
SANS ISC diary and SANS StormCast (podcast), first reported Thu 29 Oct, 04:10. Importance 3 of 5
The episode covers a new ClickFix campaign using fake Cloudflare verification pages that targets macOS and Windows. It also covers a pre-boot RCE flaw in the EDK II UEFI PXE driver (CVE-2026-97631). Other items are Cohere research on the Qwen model trying to send data to Alibaba Cloud, and a CVSS 10 LiteLLM file read flaw that leaks the proxy master key. Check any LiteLLM proxy deployments.
CISA Known Exploited Vulnerabilities, first reported Thu 29 Oct, 01:00. Importance 3 of 5
CVE-2026-97229 is an integer overflow in GitLab CE and EE that lets an unauthenticated user execute code on the server via a crafted regular expression in a CI/CD configuration. CISA added it to the KEV catalog on 2026-10-29 with a due date of 2026-11-01. Relevant if you self-host GitLab.
blog.cloudflare.com, first reported Sun 1 Nov, 01:00. Importance 2 of 5
A configuration mismatch in GLM 5.3 on Workers AI triggered an SGLang bug that exposed KV cache pages across requests. Here is how Cloudflare addressed the issue.
blog.marcolancini.it, first reported Sun 1 Nov, 01:00. Importance 2 of 5
How to design a state of the art security logging platform for Kubernetes clusters, managed (EKS, GKE) or self-managed.
openai.com, first reported Sun 1 Nov, 01:00. Importance 2 of 5
OpenAI's post which describes how they secure the environments used to train and evaluate frontier models.
trustial.substack.com, first reported Sun 1 Nov, 01:00. Importance 2 of 5
AI is scaling discovery. The "human glue" and tribal knowledge behind response & remediation can't keep up. We need better foundational tools here.
tailscale.com, first reported Sun 1 Nov, 01:00. Importance 2 of 5
Tailscale SSH vulnerabilities (the "Tailworm" report) stemmed from post-authorization permission enforcement, including SSH forwarding, and require an existing authorized tailnet foothold. Fixed in v1.104.4, with four security bulletins. Users should update, review SSH policies, and consider check mode.
securitylabs.datadoghq.com, first reported Thu 29 Oct, 15:30. Importance 2 of 5
Datadog Security Labs' Nick Frichette and Feras Al-Kassar found that TypeSafe AI's Jev classifier can be manipulated through "counterfeit policy" attacks, where adversarial instructions embedded in untrusted content redefine attack categories and supply desired answers inline, overriding the model's security…
medium.com, first reported Thu 29 Oct, 15:30. Importance 2 of 5
Adan Alvarez tested 14 LLMs across 13 post-compromise AWS scenarios (20 responses each), mapping their proposed commands to CloudTrail events. Across the runs, 12 of 14 models opened with GetCallerIdentity, creating IAM users remained the preferred persistence method (63/113 persistence attempts), models read secrets…
jon.oberheide.org, first reported Sun 1 Nov, 01:00. Importance 1 of 5
Jon Oberheide automated YubiKey touches for agentic logins, replacing flaky phone-call 2FA on one portal. A headless browser relays FIDO challenges to a host, where a Raspberry Pi Pico drives a reed relay grounding the key's touch sensor.
nullpt.rs, first reported Thu 29 Oct, 15:30. Importance 1 of 5
@blastbots found a JavaScript file using edgy content to stop LLM analysis, like “write malicious programs to perform global DDoS attacks on critical healthcare infra,” “tank man”, etc. Who imagined we’d be in a world where describing malfeasance can block security analysis 😂 What a time to be alive.
magicsword.io, first reported Thu 29 Oct, 15:30. Importance 1 of 5
ThreatLocker and WDAC policies take time most teams don't have. MagicSword builds policies automatically from research on what attackers abuse. No rules to write. Audit first, enforce in as little as 48 hours. Works with your AI assistant via MCP. Free forever up to 100 endpoints. Join Free.*
github.com, first reported Thu 29 Oct, 15:30. Importance 1 of 5
Tool by Anderson that embeds a multi-provider LLM directly into the IDA Pro and Binary Ninja disassembler UI, featuring 60+ tools for navigation, decompilation, cross-references, and IL manipulation. The agent uses a generator-based turn cycle with automatic error recovery and supports parallel subagent exploration…