Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Real-Time AI Security Monitoring: Why One Assessment Expires

A penetration test on a web application stays broadly valid until someone changes the application. An assessment of an AI system starts expiring immediately, because the system changes without anyone at your organization touching it. The same prompt can return a different answer tomorrow, and the provider can revise the model underneath you without notice. ‍

How to Connect WordPress to ChatGPT Using MCP Server | Secure MCP Server

The Secure MCP Server turns your WordPress website into an MCP (Model Context Protocol) server, allowing ChatGPT and other compatible AI clients to securely interact with WordPress. ���� �������� ����������, ������'���� ����������: • How to install the Secure MCP Server plugin on WordPress• How to configure the MCP Server• How to connect your WordPress website to ChatGPT• How to authorize ChatGPT to access WordPress• How to test the MCP connection• How ChatGPT can interact with your WordPress website through MCP• How to view and monitor audit logs of MCP activity.

Evaluating AI systems at Corelight

AI system evaluation is the process of continuously assessing AI system capabilities, limitations, and performance through quantitative and qualitative measures. Across the system lifecycle, evals provide continuous assurance: Validating system behavior before deployment and detecting drift, bias, and reliability issues in production.

How to stop sensitive data leaking into ChatGPT, Copilot and other GenAI tools

AI productivity tools are creating a prompt-level data leakage problem: 77% of employees paste data into generative AI tools, and 82% of that activity comes from unmanaged accounts, according to the LayerX Enterprise AI and SaaS Data Security Report 2025. Every paste into ChatGPT, Copilot or another GenAI tool is a potential exposure of sensitive data that traditional file-focused controls were never built to see.

Whos watching your AI A security leader panel on runtime visibility and accountability

AI is making decisions across your environment — calling APIs, accessing data, and taking action. Most organizations know it's happening, but far fewer can see it, let alone stop it. In this panel discussion, security leaders will share what they've learned deploying and governing AI workloads at scale on AWS, and what it takes to close the gap between deployment and accountability.

Episode 21 - Building AI Harnesses to Unify Detection and Response

Corelight Senior Security Engineer Jordan Hair joins Richard Bejtlich to break down how defense teams can leverage agentic AI harnesses to transform traditional security operations. By wrapping deterministic code around large language models, Hare created automated agents for alert triage, threat hunting, and detection engineering that shrink routine investigations from 45 minutes down to seconds.

American Cyber Mercenaries - The 443 Podcast - Episode 383

This week on the podcast, we discuss a new White House memorandum that creates a program to authorize American private companies to begin conducting offensive cyber operations. Before that, we discuss a vulnerability write up for a Citrix Netscaler flaw before covering yet another prompt injection vulnerability in a popular AI tool.

The risk of using abandoned packages in the age of LLMs

This post is an unfortunate affirmation of our prior research into abandoned open-source packages, where we found that 11% of the most-downloaded packages have been abandoned and not actively maintained, becoming invisible vulnerabilities to your scanner. Today we share a zip-slip vulnerability we found in extract-zip (CVE-2026-19693), an npm package with over 20 million weekly downloads.

Solving the SOC Data Problem: How Modern SIEM Platforms Cut Noise Without Cutting Visibility

Security teams have a data problem, not a detection problem. Most SOCs today aren't short on logs - they're drowning in them. Every firewall, endpoint, identity provider, and cloud workload generates a steady stream of events, and somewhere inside that noise sits the handful of signals that actually matter. The challenge isn't collecting more data. It's finding the right data fast enough to act on it.