Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Somewhere in the last few months, the math of application security quietly broke. Anthropic’s Claude Mythos Preview didn’t just analyze code, it found a 27-year-old vulnerability in OpenBSD, a 16-year-old bug in FFmpeg, and a 17-year-old remote code execution flaw in FreeBSD, entirely on its own. Then it went further: it built working exploits for them. No human guidance. No months of manual research. And by Anthropic’s own account, this is only a preview of what’s coming.

To write an AI usage policy, define its scope, list approved and prohibited tools, name the data types banned from prompts, write clear rules, align to recognized frameworks and set enforcement. A written policy is the decision layer; pairing it with technical controls is what turns those decisions into protection.

Every enterprise control framework assumes the entity that requests an action and the entity that approves it are different. Multi-agent workflows quietly dissolve that assumption. Three agents each holding modest, individually reasonable permissions can compose an action none of them was authorized to take, and no single permission grant looks wrong in a review. ‍ That is the distinguishing property of multi-agent systems rather than a harder version of single-agent risk.

Organizations preparing for the EU AI Act keep asking which standard to certify against, and the honest answer is that the ones that will matter are not finished. No harmonized standard has been cited in the Official Journal, and nothing available today confers presumption of conformity with the Act's requirements for high-risk systems. ‍

Route leaks push traffic down paths it was never meant to take. We have written and spoken publicly in the past about route leaks in Border Gateway Protocol (BGP), depicting these events as impactful incidents that cause misdirection of traffic through unintended network paths. BGP routing is driven by the relationships between Autonomous Systems (ASes), i.e., customer-provider and peer-peer.

Have you ever tried to answer “Is this container safe to run?” If yes, then you know how tricky that question is. Securing containers is one of the toughest jobs in security, and hunting vulnerabilities across an estate of them gives security folks the same vibe as hunting in the fifth domain. Container vulnerabilities can live almost anywhere, from the image layers down to the kernel, and every week, a new base image, a new dependency, or a new manifest change opens a fresh gap.