"...Sign in with ANY password: How we used AI to break into a popular chat application, ..."
Basically a brute-force attack?.. Stupidly easy to launch with AI now...
@oldguycrusty @GitHubSecurityLab not brute force; kind of a subtle javascript promises usage bug https://github.blog/security/how-to-scan-for-vulnerabilities-with-github-security-labs-open-source-ai-powered-framework/#h-signing I'm pretty sure this is the fix https://github.com/RocketChat/Rocket.Chat/pull/38118/files#diff-02105acda1e5f89d9d8350266e9323bc8a92566fd96b91905bddd3b3af8e929bR18 ... the login with any password was to a supporting microservice not the main app, presumably that's how it had not been noticed in at least the 4 years since the microservice was split out (based on skimming git blame).
(I just happened to have read the post, may've misunderstood myself, take with grain of salt!)