Ep1: Shoshana Cox | AI cybersecurity is critically different
Failed to add items
Add to basket failed.
Add to wishlist failed.
Remove from wishlist failed.
Adding to library failed
Follow podcast failed
Unfollow podcast failed
Ep1: Shoshana Cox | AI cybersecurity is critically different
-
Narrated by:
-
By:
Can you secure your AI systems?
Learn from AI cybersecurity expert Shoshana Cox why AI security is fundamentally different from traditional systems, why it’s difficult to do, and why most AI security solutions and consultancy sold today is worthless.
Shoshana is a mathematician, engineer, and former hacker working in AI cybersecurity in the national security sphere. She has written pioneering papers laying out why AI systems are so difficult to secure (her mathematical proof was recently confirmed by NIST in the US) and was part of the team that wrote the EU AI Act’s technical standards. She is a strong believer in AI and a fierce critic of companies selling security services which deliver no value!
Timestamps
00:30 Introduction
03:02 How is AI cybersecurity different from traditional cybersecurity?
05:28 In AI, data is the attack vector
08:07 Threat modelling in AI security
10:59 The challenges of implementing AI in organisations
14:41 Is current AI hype dangerous?
18:50 How should large organisations approach AI?
21:52 Internal vs external AI models
24:28 Data, process quality and AI reliability
27:23 The importance of real engineering in AI deployment
30:11 AI red teaming - as it’s being sold - doesn’t work