Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Identity threat detection and response (ITDR) tools close the visibility gap that EDR and MFA leave open. They surface credential misuse, lateral movement, and Active Directory activity that appears legitimate to endpoint and perimeter defenses. The right fit depends on your identity infrastructure, detection depth, and whether you need real-time blocking or post-event response.

Cato CTRL has discovered high-severity vulnerabilities in NVIDIA NeMo (CVE-2025-33236 with a CVSS score of 7.8) and Meta PyTorch that turns AI model files into remote code execution (RCE) vectors. The NeMo vulnerability allows RCE by importing a malicious AI model. The NeMo framework silently executes threat actor-controlled code with no warning.

Every security vendor shipping an AI product in 2026 makes the same promises. Faster triage. Shorter response times. Fewer false positives. Reclaimed analyst hours. But, six months after deployment, most security leaders still cannot answer a straightforward question from the board: Is this thing actually working?

Most organizations have a reasonable handle on their sanctioned SaaS apps. Model Context Protocol - hit 10,000 public servers within a year of launch, with 97 million monthly SDK downloads. None of those numbers capture the servers your developers configured locally. Those don't appear in any registry. They were added at the IDE level, one developer at a time, with no approval step and nothing that touches a central system. That's the inventory problem. It comes before any question of enforcement.

AI agents fall into five foundational categories: simple reflex, model-based reflex, goal-based, utility-based, and learning agents. Each is defined by how much environmental awareness and decision-making complexity the system can handle, from fixed condition-action rules to feedback-driven self-improvement.

A patch management policy is a formal organizational document that defines how software updates are identified, prioritized, tested, and deployed across an organization's systems to address security vulnerabilities that attackers commonly exploit to gain initial access.

This post covers a supply chain attack introducing a capability not seen before: using an AI coding assistant’s own GitHub access to commit malicious code to a corporate repository.

Most security teams treat a high volume of false positives as an analyst problem. Too many alerts, too little time, not enough headcount. So they add analysts, tune a few policies, and move on. That response is understandable, but it misdiagnoses the problem. When data loss prevention (DLP) false positive rates stay high over time, the issue is not a staffing gap. It is a detection accuracy problem, one that sits inside the tool, not the team.

Why do access control challenges exist, despite most companies following it? The gaps could be due to inconsistent permissions, accumulation of accesses, or poor management of user lifecycles. Access control is about governance. It answers two questions: “Who are you?” and “What are you allowed to do?” To add on, in today’s multi-cloud hybrid reality, governance is hard to handle. This isn’t another theoretical deep dive.

As AI systems become more capable and increasingly embedded into business operations, security teams are confronting a familiar challenge in a new form: speed without context. Vulnerability discovery is accelerating toward machine scale, while adversaries continue to adapt in real time. In response, the industry has gravitated toward data‑driven scoring models to help determine what deserves attention first.