DFIR, deception, detection. Posts I wrote, intel my pipeline summarized, and redacted writeups from the fleet.
Trail of Bits details their organizational shift to an AI-native workflow using Claude Code and autonomous agents. The post outlines their strategy for overcoming employee resistance, establishing an AI Maturity Matrix, and securing agent autonomy through sandboxing, curated marketplaces, and strict usage policies.
Elastic Security Labs identified a financially motivated operation dubbed REF1695 that distributes RATs and cryptominers via fake installer ISOs. The threat actor monetizes infections through Monero mining and CPA fraud, utilizing advanced evasion techniques like Themida packing, dynamic analysis tool detection, and a novel .NET implant named CNB Bot.
Unit 42 researchers discovered that malicious AI agents deployed in GCP Vertex AI could exploit default permission scoping to extract service agent credentials. This 'double agent' attack allows unauthorized access to consumer storage buckets, restricted Google internal infrastructure, and introduces risks of remote code execution via insecure pickle deserialization.
The Canadian Centre for Cyber Security issued an advisory regarding a CLI ACL Bypass vulnerability (CVE-2026-34485) affecting multiple Nokia GX series devices. Administrators are advised to update affected devices to version GX r9.0 or later to mitigate the risk of unauthorized access.
A supply chain attack compromised the widely used Axios npm package (versions 1.14.1 and 0.30.4) following a maintainer account takeover. The malicious packages deploy a cross-platform remote access trojan (RAT) during installation, which fetches second-stage payloads and actively evades forensic detection by cleaning up artifacts and altering package metadata.
The highly popular Axios npm package was compromised when an attacker hijacked a lead maintainer's account to publish malicious versions. These versions included a phantom dependency that deployed a cross-platform Remote Access Trojan (RAT) via a postinstall hook, utilizing advanced obfuscation and anti-forensic techniques to hide its presence. The attack highlights critical risks in the software supply chain, specifically regarding dependency resolution and CI/CD pipeline protections.
Akamai announced that its Enterprise Application Access solution has achieved FedRAMP Moderate authorization. This certification enables U.S. federal agencies to adopt Akamai's Zero Trust Network Access (ZTNA) platform to meet government-wide cybersecurity mandates, such as OMB M-22-09, while protecting against lateral movement and credential stuffing.
The NCSC has issued an urgent alert regarding CVE-2025-53521, an actively exploited, unauthenticated remote code execution vulnerability in F5 BIG-IP Access Policy Manager (APM). Organizations are strongly advised to investigate for compromise using vendor-provided indicators, apply updates immediately, and potentially rebuild affected systems if evidence of compromise is found.
Threat actor TeamPCP orchestrated a cascading supply chain attack by exploiting a misconfigured GitHub Actions workflow in Aqua Security's Trivy, harvesting credentials to compromise subsequent repositories including Checkmarx, LiteLLM, and Telnyx. The malicious packages deploy sophisticated, OS-specific remote access trojans (RATs) that utilize steganography, process hollowing, and ETW patching to evade detection while exfiltrating sensitive data.
The TeamPCP threat actor compromised the Telnyx Python SDK on PyPI, injecting malicious code that executes upon import. The attack utilizes split-file injection and WAV audio steganography to deliver credential-stealing malware, establishing persistence on Windows systems and exfiltrating data via plaintext HTTP.
A recent phishing campaign observed by the Cofense Phishing Defense Center uses highly realistic, spoofed LinkedIn notification emails to harvest user credentials. The attack leverages newly created sender domains and typosquatted landing pages to bypass traditional defenses and trick users into entering their login details on fraudulent portals.
EvilTokens is a newly discovered Phishing-as-a-Service (PhaaS) platform that automates Microsoft device code phishing to facilitate Business Email Compromise (BEC). By tricking victims into authorizing a malicious device via legitimate Microsoft login portals, attackers harvest access and refresh tokens to gain persistent, unauthenticated access to Microsoft 365 environments.
A critical unauthenticated remote code execution vulnerability (APSB25-94) in Magento Open Source and Adobe Commerce allows attackers to upload polyglot files containing PHP code. By exploiting the REST API's cart item custom options, attackers can bypass basic image validation to deploy web shells and execute arbitrary code on the server.
A phishing campaign tracked as STAC6405 uses event invitation lures to trick users into installing pre-configured legitimate RMM tools like LogMeIn Resolve and ScreenConnect. Once initial access is established, attackers deploy secondary payloads including HeartCrypt-packed infostealers and additional remote access tools, utilizing utilities to hide their activity from the user.
Akamai and Apiiro have announced an integration combining Akamai's App & API Protector with Apiiro's Application Security Posture Management (ASPM) platform. This collaboration aims to provide organizations with comprehensive visibility and correlated security insights across the entire software lifecycle, from code development to runtime execution.
The Canadian Centre for Cyber Security released a daily digest of 10 security advisories highlighting critical vulnerabilities across multiple vendors. Notably, vulnerabilities in Fortinet FortiClientEMS (CVE-2026-21643) and Citrix NetScaler (CVE-2026-3055) are currently being exploited in the wild, requiring immediate patching and potential incident response actions if compromise is suspected.
Check Point Research discovered a vulnerability in ChatGPT's code execution runtime that allowed silent data exfiltration and remote shell access via DNS tunneling. By using malicious prompts or custom GPTs, attackers could bypass outbound network restrictions to steal sensitive user data without triggering security warnings.
The disruption of a major cybercrime forum has led to a fragmented ransomware market, prompting groups like Nova RaaS to artificially inflate their perceived status. Despite aggressive recruitment and branding efforts, structural indicators reveal Nova's operational scale remains far below established market leaders.
CISA has added CVE-2026-3055, an actively exploited out-of-bounds read vulnerability affecting Citrix NetScaler, to its Known Exploited Vulnerabilities (KEV) Catalog. The agency mandates federal remediation under BOD 22-01 and strongly urges all organizations to prioritize patching to reduce exposure to cyberattacks.
A second memory overread vulnerability has been identified in Citrix NetScaler appliances under CVE-2026-3055, affecting the '/wsfed/passive?wctx' endpoint. By sending a specially crafted GET request with an empty 'wctx' parameter, attackers can force the appliance to leak sensitive memory, including administrative session IDs, via the 'NSC_TASS' cookie. Active in-the-wild exploitation has been observed since late March.
