August 6, 2024
Fortinet’s Jim Richberg is a battle-tested veteran. His opinions are sought out in Congress and all over the world. Today, we sat down with Jim to assess the implications of the federal government using AI when it comes to cybersecurity.
Jim likes to distinguish between two broad categories of AI: discriminative AI and Generative AI.
Discriminative: AI that can classify data but cannot generate it. From a cyber perspective, it can assess sentiment and image classification. It can alert when there is a potential threat.
Generative AI: produces results based on amalgamation of existing data. This can be text, music, and even designing images. There is a possibility that a Generative AI application can look at a network and generate a report finding vulnerabilities.
"Allow you to develop code for new exploits, but the reality is they don’t always work, and they are often noisy, and not the kind of stealthy tool tan an adversary would use. "
So far, these have been defined, but the results have been few.
Jim Richberg thinks federal organizations are being limited to not having standards in contract terms as they procure AI.
One topic that many seem to overlook is the power requirement for AI. Some solutions include modifying chip design. But Jim cautions that it can take as long as five years to get a data center up and running while energy concerns are building daily.
This is a fascinating discussion that includes Jim’s delineation of MDM of Misinformation, Disinformation, and Malinformation.
If you enjoyed this episode, then you may want to listen to
Episode 177 – A New View on Threat Intelligence
Complements the cybersecurity focus by providing context on evolving threat intelligence, which is critical to global cybersecurity efforts.Episode 178 – Akamai & Identity Management
Relevant for conversations around secure digital identity frameworks—important when discussing global AI-driven cybersecurity strategies.
Leave a Reply