Windows 11 update breaks business logins

Hi there, this is your daily ☕️ Cyberpresso.

In today's Cyberpresso:

🖥️ Windows 11 update breaks business logins

🔌 Brevo hack infects 100,000 WordPress sites

🚨 Cisco patches exploited login-bypass flaw

🦠 Extension can hijack AI browser assistants

☎️ Attackers can hijack PBX servers

Plus: 💡 4 strategies & tactics, 🎁 6 other news you might like, 🧰 6 tools, and 📚 5 papers.

🖥️ Windows 11 update breaks business logins LINK

A Windows 11 security update, KB5124008, released on September 8, 2026, is breaking Active Directory domain trust on some enterprise machines, leaving users unable to sign in even with valid credentials on versions 25H2 and 24H2.

The problem is tied to Machine Identity Isolation, a Credential Guard feature in enforcement mode that moves the machine secret into Credential Guard; if it cannot complete machine authentication after a restart, the secure channel breaks and interactive logon fails.

Microsoft has not confirmed the cause or issued an official fix, but administrators stopped the failures by disabling Machine Identity Isolation and forcing its registry value to 0, then resetting the machine password, though some systems still required a full domain rejoin.

🔌 Brevo hack infects 100,000 WordPress sites LINK

A breach of Brevo's infrastructure on September 14 pushed malicious JavaScript through the company's hosted widgets and scripts to more than 100,000 websites, including operators embedding its tracking SDK or chat components without changing anything themselves.

Modified copies of Brevo's sdk-loader.js and brevo-conversations.js loaded attacker code from subdomains under sendibt1.com; for logged-in WordPress admins it used their active session to upload a plugin from cdn10.sendibt1.com/p/wm.zip, suspected to plant persistent backdoor access.

Unauthenticated visitors got a full-screen ClickFix "verify you are human" prompt tricking them into pasting and running a malicious command; the malicious hosts stopped resolving September 15, Brevo restored the files, and admins should review September 14 plugin-upload logs.

🚨 Cisco patches exploited login-bypass flaw LINK

Cisco has issued emergency patches for a login-bypass flaw in its Identity Services Engine (ISE) that attackers have already been exploiting in the wild before any fix was available.

The maximum-severity bug (CVE-2026-76460, CVSS 10/10) lets an attacker send crafted requests to an API endpoint that fails to enforce authentication, bypassing the web management interface to reach the device with no login needed.

Successful exploitation runs commands with root privileges, letting attackers hide or delete traces; Cisco says to upgrade ISE or ISE-PIC to fixed patches, check 'access.log' for suspicious usernames, and re-image compromised nodes.

🦠 Extension can hijack AI browser assistants LINK

A proof-of-concept attack called BragJack lets a malicious browser extension hijack the communication channel to built-in AI agents in five agentic browsers, Chrome with Gemini, Edge, Opera Neon, Perplexity Comet, and Claude in Chrome, and force them to obey attacker commands.

Rather than hiding instructions in content, the attack abuses a shared design flaw letting an untrusted extension cross the boundary meant to isolate it from the privileged AI agent, so attackers can repeatedly push prompts until the agent takes screenshots, reads local files, or activates the camera and microphone without user clicks.

Researcher Gal Weizman of Forever Security reported it to all five vendors, who resolved the issues; Google and Microsoft issued CVE-2026-0628 and CVE-2026-55945 for Chrome and Edge, and he advises keeping Chromium-based browsers updated and removing unvetted extensions.

☎️ Attackers can hijack PBX servers LINK

Attackers are actively hijacking internet-facing Issabel PBX servers by exploiting a critical flaw (CVE-2026-89026, CVSS 9.3) in the Issabel Framework that lets unauthenticated remote attackers run operating-system commands on vulnerable phone systems.

The bug exists because the framework shipped a hard-coded JWT signing key, so an attacker with no account or password can forge a legitimate-looking bearer token and hit the originate API endpoint to make Asterisk run arbitrary commands.

Shadowserver first saw exploitation on September 9, 2026; administrators should update to the version containing the fix (commit b97dbaf0), confirm the shared key is gone, restrict PBX and API access, and review logs for suspicious originate requests.

💡 Strategies & Tactics

> GitHub Actions Adds cache-mode to Limit Cache Poisoning Risk: GitHub Actions now lets maintainers grant each workflow only the cache access it needs, blocking attackers from planting malicious code trusted jobs later run.

> CISA Releases Guidance on Deploying Cyber Decoys: Deploy fake systems that look real to lure and detect attackers early, since intruders using stolen credentials often evade normal defenses.

> How I Tricked OpenClaw Into Attacking Its Own Network: A NAT64 SSRF Bypass: A crafted IPv6 address fooled OpenClaw's URL-fetch guard into reading harmless decoy bytes while routers sent traffic to internal servers, enabling attacks on private networks.

> From Bug to Schema: Exploring Error-Based SQL Injection on an Authenticating Portal: Prove a critical database flaw by leaking only its structure names, not real data, staying within a bug bounty program's legal limits.

Other news you might like

• GhostCode Abuses Microsoft Entra Device Enrollment to Maintain Access After Token RevocationLINK

• U.S. CISA adds Acronis Backup, Cisco ISE, and Google Pixel flaws to its Known Exploited Vulnerabilities catalogLINK

• NightEagle targets Russian companiesLINK

• New Android Malware Steals Banking PINs and Reinstalls Itself After Users Delete ItLINK

• US takes down NightmareStresser DDoS-for-hire platformLINK

• Chinese hackers use SparroWocky malware in govt espionage attacksLINK

🧰 Trending tools

Halo: an API-first platform that detects deepfakes and synthetic media across text, image, and audio, helping fraud and trust teams block attacksLINK

MonoCloud for Startups: manages authentication and Cedar-based authorization across users, APIs, and AI agents, letting you control, audit, and revoke access, free for a yearLINK

Execlave: enforces runtime policies, kill switches, and audit logs on autonomous AI agents in under 20ms, mapping to SOC 2, EU AI Act, and ISO 27001 complianceLINK

Aegisora: an open-source proxy that secures LLM agents with least-privilege API access, PII masking, prompt-injection blocking, and audit logging for production deployments.LINK

Agent Beacon is the world's first open-source telemetry layer for AI agents wherever they run: locally, in CI, in the browser, or in the cloud.LINK

Linux Doctor: read-only diagnostic tool that runs system health checks and explains detected problems without making changes to the machine.LINK

📚 Trending papers & reports

Personalized security chatbots tailored to a user's ongoing conversation make people rate cybersecurity advice as more helpful and more likely to be followed, based on a 7-day test with 57 participants and over 1,000 questions.LINK

Onboard drone security spots cyberattacks and malfunctions in real time by watching a drone's sensor and command data as it flies, running directly on the drone despite its limited battery and computing power.LINK

Encrypted-traffic leakage accounting breaks down exactly how much a website visit leaks through packet size, direction, and timing, showing that popular defenses like FRONT cut direction leaks ~63% but leave timing exposed.LINK

Reliable learners are a training method that lets prediction systems flag when they might be wrong under hostile inputs or shifted data, giving provably correct answers in messy real-world conditions where accuracy guarantees usually break down.LINK

Automated bug-hunting hints use AI to write the expert guidance that helps security testing tools find software flaws faster, matching human experts and removing the need for scarce specialists, though the speed gains proved inconsistent.LINK

See you tomorrow for a new dose of ☕️ Cyberpresso!