Anthropic and OpenAI missed 6 curl vulnerabilities found by Aisle
Aisle, an AI-driven cybersecurity startup, just discovered six low-severity vulnerabilities in curl, a tool many developers use to transfer files online.
What's impressive? Anthropic's Mythos and OpenAI Codex Security totally missed these bugs.
The curl team confirmed the issues and patched them fast in version 8.22.0, released on September 2, 2026.
Experts praise Aisle's system over model
Instead of relying just on raw AI power, Aisle uses a "system over model" approach, basically focusing on smarter workflows and understanding how code fits together.
Experts like Greg Kroah-Hartman, the maintainer of the Linux stable kernel, and Jim Fuller, a Red Hat senior principal software engineer specializing in security, praised Aisle's method for actually spotting real risks in big open-source projects, making the internet a little safer for everyone who depends on these tools.