OAuth Tokens Taken, Vulns in Medical IoT, Scoring a Proactive Security Culture – ASW #193

Valid credentials are the best backdoor into any system. Attackers managed to obtain OAuth tokens and abuse that access to download repo information -- likely searching for even more instances of hard-coded secrets. GitHub also discussed a security issue in Git that primarily affects Windows users, which is unrelated to the stolen token issue. It's not exactly a path traversal flaw in the sense of breaking out of a subdirectory, but it does touch on nuances of traversing directories and desired behavior when ownership changes. Check it at out https://github.blog/2022-04-12-git-security-vulnerability-announced/

If you like crypto (the useful math kind that solves problems), then you'll enjoy the deep dives of this article and the series that follows. For others, even if the details seem incomprehensible, the article raises some good points that apply more generally to application development. For example, one of the observations on how these flaws occur "comes down to a combination of ambiguous descriptions in academic papers and a general lack of guidance around these protocols." The easy summary of that is "needs more documentation", but the more effective summary is "needs clear communication for developers" -- and that's a topic we revisit on lots of episodes.

Ah, medical droids -- 21-B fixed up Luke in "Empire Strikes Back", GERTY kept an eye on Sam in "Moon", and one of my recent favorites is David from "Prometheus" (and Weyland's advertisement for that series of android, https://youtu.be/qgJs7uluwlU). But closer to 2022, IoT in the medical space still has flaws related to XSS, passwords stored as MD5 hashes (!?), and an overall lack of strong authentication. Kudos to Cynerio for a nice write-up that explains the flaws without overhyping them *and* provides a more detailed PDF without requiring a signup to a marketing list first. Check it out at https://www.cynerio.com/blog/cynerio-discovers-and-discloses-jekyllbot-5-a-series-of-critical-zero-day-vulnerabilities-allowing-attackers-to-remotely-control-hospital-robots

ASN parsing is notoriously prone to error and abuse that leads to security issues. This article actually covers a flaw from 2021, but its appeal is in the type of questions it poses rather than merely a technical review of the flawed code. One question is whether or how fuzzing could have found this issue. Another question relates to the wisdom of forking code and the kinds of flaws or missed patches that can creep into forks, even well-maintained ones.

This flaw and attack falls into the "dead simple" category and essentially boils down the executing attacker's JavaScript by clicking on a link. But the more interesting way to talk about this is security design and UX that helps users understand the consequences of actions or, even better, avoids consequences altogether. This security and UX failure hearkens back to papers like "Why Johnny Can't Encrypt" about GPG and the lock icon in browsers. Designing security controls and information for other security subject matter experts doesn't help users.

This article walks through the creation of metrics that inform teams about a security topic followed by a set of concrete actions those teams can take to improve their scores against the metrics. It's nice to see a team talking about metrics in a way that keeps them simple and ties them to things a team can influence.

OpenSSH is switching to "post-quantum" Streamlined NTRU Prime + x25519 key exchange, in hopes to prevent attackers who sniff/store data now, with hopes to decrypt at some point in the future when quantum computers realize the ability to decrypt public key cryptography. More info on NTRU Prime is available at https://ntruprime.cr.yp.to/

The interesting part of offering complex open source software as a service is how to you ensure that every aspect of it has been properly secured?