Nov 10, 2025

The cyber frontlines are shifting fast, Digital Defenders.

AI isn’t just assisting — it’s under attack. From hijacked plugins to zero-click exploits and stealthy malware, the tools built to empower us are becoming threat vectors.

Here’s what you need to know:

Zero-Day Download

AI Engine Plugin Flaw Puts 100K WordPress Sites at Risk

A vulnerability in the AI Engine plugin exposes 100,000+ WordPress sites to privilege escalation.

CVE-2025-11749, linked to the plugin’s No-Auth URL feature, leaks bearer tokens via the /wp-json/ API, letting attackers seize admin control.

AI Engine Plugin Flaw Puts 100K WordPress Sites at Risk

Source: Souvik Banerjee/Unsplash

A patch has been released for vulnerability.

Update AI Engine to version 3.1.4 immediately, rotate all existing bearer tokens, disable unused API endpoints, and deploy a web application firewall to block unauthorized requests.

Tenable Uncovers ChatGPT Data Leak Risk

Tenable researchers found ChatGPT flaws enabling indirect prompt injection, data theft, and persistent cross-session attacks.

Attackers can exploit weak isolation between ChatGPT and SearchGPT to inject hidden commands via web content, stealing data without user awareness.

This turns browsing and memory features into exfiltration tools.

Disable or strictly limit AI browsing and memory integrations, monitor for unusual output behaviors, and treat conversational context as sensitive data requiring zero-trust controls.

Claude Desktop Exploit Enables Full System Takeover

Koi researchers found RCE flaws in Claude Desktop extensions for Chrome, iMessage, and Notes, allowing simple prompts to trigger full system takeover.

Unsanitized input in AppleScript calls allowed crafted queries to run arbitrary code and exfiltrate SSH keys, credentials, or stored passwords. Patches have been released for the vulnerability.

Update Claude Desktop extensions, isolate AI tools from sensitive files, and audit extension permissions.

Share this newsletter!

Pulse Check

How confident are you in spotting AI security risks?

🔒 Very — I assess AI tools like any other attack surface

⚙️ Somewhat — I review settings but might miss deeper risks

🤷‍♂️ Not much — I trust vendors or defaults to handle security

Yesterday's Pulse Check Results

How would you rate your own reversing chops — honestly?

Security Tools You Can Use

Don’t wait for the next breach to wake you up — lock down your systems before attackers even try.
Turn the tables on cybercriminals — master their moves and cash in on bug bounty rewards while you can.
No degree? No problem. Microsoft’s Cybersecurity Analyst training puts you on the fast track to a cybersecurity career.

🛡️ This section contains sponsored cybersecurity insights. Advertise with us!

Threat Tracker

Android Zero-Click Exploit Enables Silent Device Takeover

A recently disclosed zero-click vulnerability in Android’s System component allows attackers to execute code and seize full device control.

CVE-2025-48593 affects multiple Android versions, potentially exposing billions to data theft and ransomware amid ongoing patch delays across the mobile provider.

Android Zero-Click Exploit Enables Silent Device Takeover

Source: Guido Coppa/Unsplash

Google has already released a patch for the vulnerability.

Apply the latest patch, enable automatic system and Play Protect updates, and enforce enterprise patch compliance through MDM policies to prevent silent compromise.

Malware Abuses OpenAI API for Covert Command Operations

Microsoft researchers uncovered SesameOp, a stealthy espionage backdoor that exploits OpenAI’s Assistants API for covert C2 operations.

The malware hides within legitimate OpenAI traffic, using its API for covert persistence and data theft. This backdoor is being actively used in the wild.

Restrict API usage to verified keys and approved domains, monitor outbound connections for anomalies, and enforce zero-trust controls around AI service integrations to detect misuse early.

Security Tip of the Day

Could AI Tools Be Your Next Attack Vector?

As AI tools become part of daily workflows, endpoints are becoming prime targets for data theft, prompt injection, and model manipulation.

Attackers exploit AI integrations in apps and agents for system access — turning convenience into compromise. Strong endpoint defense is critical.

Could AI Tools Be Your Next Attack Vector?

Source: Steve Johnson/Unsplash

How to secure endpoints running AI tools:

Limit AI access and permissions: Restrict file, network, and credential access, and allow only approved AI tools or extensions to reduce exposure.
Monitor and contain AI activity: Use EDR solutions to detect abnormal behavior and isolate AI workloads in sandboxes to prevent lateral movement.
Secure and educate: Enforce strict API management, revoke unused tokens, and train users to recognize AI-related threats like prompt injection or data leakage.

AI-driven productivity shouldn’t come at the cost of security. By hardening endpoints and managing AI access thoughtfully, organizations can harness innovation without exposing critical assets.

Meet Our Writer:

Ken Underhill

Ken Underhill is an award-winning cybersecurity professional, bestselling author, and seasoned IT professional. He holds a graduate degree in cybersecurity and information assurance from Western Governors University, bringing years of hands-on experience to the field.

Secure your spotlight and Advertise in Cybersecurity Insider!

Cybersecurity Insider is a TechnologyAdvice business

TechnologyAdvice, 3343 Perimeter Hill Dr., Suite 215, Nashville, TN 37211, USA.

Want to change how you receive these emails? Update your preferences or unsubscribe here.