Security Intel Feed

Cyber Hose

Page 7 of 44

Escape found stored XSS in two AI chatboxes via Markdown rendering — Escape's AI pentesting agent induced chat models to emit malicious Markdown that executed JavaScript in users' browsers across two unrelated products.
Peer2Profit Turns Employee Devices Into Gateways to Internal Networks — Silent Push found 117,224 AstroProxy IPs including residential nodes that exposed internal router interfaces through enrolled proxy devices.
CVE-2026-77710 — STIX2 Parser Confusion and Mass Assignment Allow Unauthorized MISP Attribute — A CVSS 6.9 vulnerability in MISP allows unauthorized attribute creation via STIX2 parser confusion and mass assignment.

CVE-2026-18409 — WPForms Pro through 2.0.0.2 - Unauthenticated Stored Cross-Site Scripting — A CVSS 7.2 unauthenticated stored XSS vulnerability affects WPForms Pro versions through 2.0.0.2.

Critical Microsoft Entra ID RCE flaw exploited in the wild — CVE-2026-69836 is a critical remote-code-execution flaw in Microsoft Entra ID that Microsoft says is being exploited in the wild.

SPIP Before 4.4.20 Hit by Unauthenticated RCE CVE-2026-77647 — CVE-2026-77647 allows unauthenticated remote attackers to execute arbitrary code on SPIP installations before version 4.4.20 and was exploited in the wild.
SynkLoader malware uses fake Windows lock screens to steal passwords — SynkLoader deploys a convincing fake Windows lock screen to harvest user login passwords using in-memory modules across multiple languages.
CVE-2026-72843 — EverShop Critical Unauthenticated Account Takeover — A CVSS 9.3 vulnerability enables unauthenticated attackers to seize EverShop user accounts without credentials.

Attackers Poisoned Three Rust Crates With Build-Time Malware — A supply-chain attack injected build-time malware into three Rust crates, potentially compromising downstream projects that depend on them.
CVE-2026-71485 — Centrifugo: Client-forgeable headers emulation lets any client spoof headers — A CVSS 9.1 vulnerability in Centrifugo allows any client to spoof headers through client-forgeable header emulation.
Jake Williams Releases CUSTODY Framework to Constrain AI Agents — The CUSTODY framework aims to constrain enterprise AI agents operating inside networks following attacks involving OpenAI and Hugging Face.

AI-generated scripts target internet-exposed Siemens S7 PLCs — Attackers are using AI-generated tools disguised as OT monitoring software to read and write PLC memory across water, energy, and manufacturing sectors.
Suspected Russian Hackers Abuse Google OAuth and WhatsApp to Hijack Accounts — Threat clusters UNC6293, UNC7005, and UNC5976 are abusing legitimate Google authentication flows and WhatsApp device-linking for cyber espionage.
CVE-2026-19586 — CVSS 9.3 — Pre-Authentication OS Command Injection in Omada Gateways on OpenVPN Server — A pre-authentication OS command injection vulnerability in Omada Gateways allows unauthenticated remote attackers to execute commands via the OpenVPN server.

CISA Flags Exploited TrueConf Server Flaws Used in Head Mare Attacks — Head Mare exploits unpatched TrueConf servers to deploy PhantomCore and PhantomGraph backdoors, impacting many organizations.
TeamPCP LiteLLM attack exposes credentials from nearly 2,500 organizations — A supply-chain attack on LiteLLM leaked credentials from thousands of organizations including major tech companies.
Researcher registers Linux device to Apple Find My People — A researcher accessed Apple Find My People location shares from a Linux device using advanced authentication techniques.
N-able Passportal bug exposes password-vault master keys — A bug in N-able Passportal exposed password-vault master keys, affecting managed service providers and small businesses.

Critical isolated-vm flaw enables sandbox escape and potential host RCE — A type confusion vulnerability in isolated-vm allows attackers to escape sandboxes and hijack host control flow.
Transparent Tribe Targets Afghan Telecom With PATCHCORD Backdoor — APT36 targets Afghan telecom and South Asian infrastructure using new backdoors and sophisticated C2 techniques.
Encrypted web-page prompts can exfiltrate Grok users’ chat data — Malicious encrypted prompts can steal Grok chatbot user data including names and locations.
Phishing toolkit covertly enrolls hidden passkeys on compromised Google accounts — A phishing toolkit secretly registers attacker-controlled passkeys on Google accounts after compromise.

UAT-10147 deploys SPECTRE cross-platform backdoor with rootkit and BYOVD — Cisco Talos has identified SPECTRE, a cross-platform backdoor used by UAT-10147.
Operation CameraSwarm Compromised 14,530 Dahua IP Cameras — Attackers compromised 14,530 Dahua IP cameras in Operation CameraSwarm.
ClawHavoc Poisoned OpenClaw’s ClawHub With Malicious Agent Skills — ClawHavoc used poisoned OpenClaw skills to deliver information-stealing malware.
Alation confirms cyberattack involving unauthorized access to its systems — Alation confirmed a cyberattack involving unauthorized activity in one system.
ClarityCheck exposed 9 million facial images in unsecured cloud storage — ClarityCheck exposed more than nine million facial image files in an unsecured cloud database.

Attackers Exploit Critical MLflow SSRF Flaw to Steal Cloud Credentials — CVE-2026-64849 in MLflow versions before 3.15.0 allows unauthenticated SSRF to reach cloud metadata endpoints and steal credentials.
Critical Zimbra RCE Vulnerability Is Being Actively Exploited — CVE-2026-73570 enables unauthenticated command execution via SNMP notification processing in Zimbra Collaboration Suite before 10.1.20.
ToxicPanda 2.0 Expands Android Banking Fraud With On-Device Attacks — The trojan targets 349 financial applications across 16 countries using overlays, Accessibility Services, and 167 remote commands.
CyberPanel Pre-Auth RCE Chain Enables Shell Access on Hosting Servers — CVE-2026-41473 and CVE-2026-41472 in CyberPanel before 2.4.4 allow unauthenticated database writes and stored XSS leading to OS command execution.

Assess Your Exposure

Start with the free Posture Self-Check to see where you stand against the current threat landscape.