Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

We studied what happens when Large Language Models (LLMs) generate vulnerability patches for recently disclosed, complex vulnerabilities. Our data shows that LLMs produce Fix-Like Artifacts with Embedded Defects (FLAWED) 53.9% of the time when complex patches are required.

AI has changed the calculus of a credential attack. Before, finding and exploiting credentials in an enterprise environment required time, patience, and human judgment. An attacker had to decide which accounts were worth testing and which systems were worth reaching. Many credentials never made the list.

This year National Social Engineering Day falls on Aug. 6. This day is designed to give us an opportunity to remind people that cybercriminals do not always need sophisticated malware, an undisclosed vulnerability or a dark room filled with glowing monitors, sometimes, all they need is a good story. Social engineering existed long before computers. Confidence tricks, impersonation, false authority and appeals to greed or fear have been used for centuries.

Construction firms have invested heavily in technology to connect project information. Drawings, specifications, RFIs, submittals, BIM models, photos, and field reports are increasingly accessible from anywhere, helping office and field teams work from the same information. Connecting project information is a critical first step. It improves collaboration, reduces rework, and helps office and field teams work from the same information.

Today, Cyberhaven is joining the Open Secure AI Alliance to help advance a future in which enterprises can adopt open-source AI without compromising security, compliance, or control over their data. Cyberhaven is betting on a world where companies are free to choose from many models, agent frameworks, and open harnesses, as well as run them on infrastructure they control. That choice matters. Enterprises will not standardize on a single model or AI platform.

Every enterprise is racing to adopt AI, and every security team is racing to keep up without becoming the department that says no. Most are still evaluating tooling built specifically for AI agents. But there's a faster, simpler starting point available today: the large majority of AI activity right now runs on behalf of a signed-in user, what we call “on-behalf-of” (OBO) access. Secure that identity well, and you've secured the AI acting through it.

A developer installs a skill to make their coding agent less chatty. It works. It also, the first time the agent uses it, reads the AWS credentials on that laptop and sends them to a domain no one recognizes. No one wrote obviously malicious code and no one approved a change. A file landed in a folder, the agent loaded it on the next run, and production credentials were gone.

Agentic AI security solutions help teams discover, govern, monitor, and control AI agents, copilots, LLM apps, MCP servers, and autonomous workflows. For security and DevOps leaders, they matter because agents can act across production systems. This guide compares leading tools and explains how to choose the right fit. AI agents are moving from assistants to actors.

On a Tuesday morning in Q3, a procurement agent at a mid-market manufacturer approved a $340,000 payment to a vendor account. The vendor name matched the approved-vendor list. The invoice format matched the standard template. The agent verified both, cross-checked the amount against historical purchase orders, and released the payment through the treasury API within eleven minutes of the invoice arriving. No human touched the transaction.

Harness engineering is the practice of building the layer, including code, that turns an AI model from a text generator into an agent that can take actions. In short, an AI agent is a model plus a harness. The model decides what to do next, and the harness makes it happen, connecting the model to tools, context, external systems, and validation. In a lot of practical work, and especially in security work, the harness decides the quality of the output more than the choice of model does.