In episode 81 of the We Hack Purple Podcast host Tanya Janca spoke to Diana Kelley, Chief Information Security Officer (CISO) at Protect AI. Diana and Tanya worked together at Microsoft, and to say that Diana is a pillar of the information security industry is somewhat of an understatement. Together they discussed problems with Large Language Models (LLMs) ingesting crappy code, and bad licenses, the OSSF (and it's goodness), and that sometimes people don't even realize they are breaking software licences when they use what an LLM has produced.
We discussed the fact that if a CVE comes out for a library an LLM gave you, but it didn't identify it with the correct name of the library, you wouldn't receive notifications about it. She clarified how ML pipelines are set up, how data scientists work, with insecure juniper laptops all over the place (perhaps a generalization on my part). We discussed how data science seems to be a topic a lot of CISOs are pretending aren't in their domain to protect, but both of us agreed that is not so. They have some of the most valuable data your organization can possess.
We also covered best practices for securing MLSec, the OWASP Top Ten for LLMs, and the new free community her company has started MLSECOPS. She also released an update version of her book, Practical Cyber Security Architecture!
.
Diana Links:
.
Bio: Diana Kelley is the Chief Information Security Officer (CISO) for Protect AI. She also serves on the boards of Cyber Future Foundation, WiCyS, and The Executive Women’s Forum (EWF). Diana was Cybersecurity Field CTO for Microsoft, Global Executive Security Advisor at IBM Security, GM at Symantec, VP at Burton Group (now Gartner), a Manager at KPMG, CTO and co-founder of SecurityCurve, and Chief vCISO at SaltCybersecurity.
.
Very special thanks to our sponsor!
Semgrep Supply Chain’s reachability analysis lets you ignore the 98% of false positives in open source vulnerabilities and quickly find and fix the 2% of issues that are actually reachable.
Semgrep also makes a ludicrously fast static analysis tool They have a free and paid version of this tool, which uses an open-source engine, and offers additional community created ruleset! Check out Semgrep Code HERE