Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Ask most security tools what an AI coding agent just built, and they can tell you. Ask what it was allowed to consume before it started, and far fewer have an answer. That gap, between watching agentic development and controlling it, is what securing it actually comes down to. Securing agentic development means stopping risk before it enters a build, not flagging it after. And risk prevention has a prerequisite most approaches skip: you can only account for the assets you actually hold and manage.

A developer on your team just told Claude Code to connect to a new MCP server, the protocol coding agents use to reach organizational tools and data. Nobody in security reviewed it. Nobody in security even knows it happened. For two-thirds of enterprises, the primary obstacle to scaling agentic development isn’t budget or headcount — it’s security risk.

Just last week, OpenAI and Hugging Face jointly disclosed what may be the first incident of its kind: during an internal evaluation of frontier cyber capabilities, OpenAI’s models, running deliberately without production safeguards in an isolated research environment, autonomously discovered and employed chained vulnerabilities to escape its sandbox, reach the open internet, and extract evaluation answers from Hugging Face’s infrastructure.

The JFrog Security Research team has discovered and disclosed a typosquatted NuGet package named Newtonsoftt.Json.Net. Note the double t and the.Net suffix. This package has been masquerading as the popular Newtonsoft.Json library while quietly shipping a trojanized fork. The trojan rigs Digitain, an online betting platform, and in later generations, exfiltrates rigged round results to an attacker-controlled server, utilizing the header X-Seq-ApiKey: theperfectheist2025.

What is DevGovOps? DevGovOps is a Software Supply Chain Engineering practice that integrates continuous governance and compliance into the DevOps software delivery lifecycle. Rather than treating compliance as a retrospective, manual hurdle, DevGovOps ensures that policy enforcement, continuous auditability, and cryptographic traceability are natural outputs of every release.

AI agents and LLMs are already building, analyzing, and deploying code across your software development lifecycle. As software supply chains become increasingly AI-driven, proactive security and access controls are your only path to success. To effectively govern authentication and permissions without sacrificing development speed, you must update your access management strategies.

At JFrog, our Security Research team continuously monitors and analyzes newly disclosed CVEs across the open-source ecosystem. Throughout our research, we have repeatedly observed cases where the assigned severity score does not accurately reflect a vulnerability’s real-world impact or exploitability. In fact, during 2025, JFrog researchers reassessed NVD critical-severity vulnerabilities and concluded that 96% warranted a lower severity rating.

As organizations aggressively shift from static Large Language Model (LLM) chatbots to fully dynamic, autonomous AI agents (e.g. systems designed to plan workflows, call APIs, write runtime code, and modify enterprise databases), traditional compliance and governance frameworks are hitting a breaking point. A landmark press release from Gartner highlights a critical systemic risk: treating AI agent governance as a monolithic, one-size-fits-all policy guarantees project failure.

There’s a category of security risk that most organizations aren’t ready for. It doesn’t live in your code repository, your CI pipeline, or your developer laptops. It lives in your runtime, in the autonomous AI agents already running in your environment, extending their own capabilities, and making decisions that no human explicitly approved. This is the challenge JFrog set out to address with our integration with NanoCo AI and their open-source agent framework, NanoClaw.

JFrog Security Research recently discovered and disclosed a critical vulnerability in FFmpeg, the world’s most widely deployed media processing framework. The discovered vulnerability, which we’ve named PixelSmash, is CVE-2026-8461 – a heap out-of-bounds write in the MagicYUV decoder (CVSS 8.8 High). We escalated this vulnerability from a simple crash all the way to reliable remote code execution – all it takes is processing a single malicious media file.