In the latest episode of the Razorwire podcast, I am delighted to welcome back our esteemed cybersecurity professionals, Oliver Rochford and Richard Cassidy. Today, we delve into the fascinating realm of generative AI and its applications in the cybersecurity landscape.

We kick the episode off with an overview of generative AI. We discuss how it works and its training on extensive datasets to infer statistical relationships between words and concepts. While major cybersecurity vendors such as Google, CrowdStrike, SentinelOne, and Microsoft have announced integrations with generative AI, Oliver issues a cautionary note, highlighting that its capabilities are often subject to overhype.

We discuss the accuracy of generative AI's representation in the business community. Listen in to hear our consensus: Is it possible for generative AI to live up to the advanced AI depicted in science fiction?

Delving into practical cybersecurity use cases and exploring risks associated with explainability, trustworthiness of outputs, and potential regulatory implications

The aim of this episode is to give you valuable advice for venturing into the realm of generative AI. Tune in to the Razorwire podcast for an in-depth exploration of this evolving technology.

Andrés Horowitz has said that 80% of all of the investment in the generative AI startup goes on compute costs. They worked out that one training run on GPT, I think, 3.5 costs somewhere between half a million to $3,800,000. Is it even affordable?"

Oliver Rochford

Listen to this episode on your favourite podcasting platform: https://razorwire.captivate.fm/listen

In this episode, we covered the following topics:

- Big Tech's control over the conversation and concerns about AI

- Inconsistencies in the guidelines and censorship policies of platforms like Spotify, Apple, and YouTube limit what can be discussed and criticised.

- The limitations and potential dangers of Artificial Generative Intelligence

- The different opinions and viewpoints surrounding NFT technology and its impact and significance

- Importance of not overhyping NFTs and allowing for experimentation and exploration of new use cases

- Limitations of Gen AI tools, particularly in terms of explainability, interpretability, and trustworthiness of data

- Advising caution when utilising AI tools for security purposes and the importance of trust and verification

- How AI tools can help with paralysis and confusion in data analysis

- Examining the high valuation of OpenAI and people's unrealistic expectations of AI due to Hollywood portrayals

- Exploring the potential of AI-powered language models like Chat GPT, their integration into various products, and the need to avoid false information

GUEST BIOS

Oliver Rochford

Oliver has worked in cyber security as a penetration tester, consultant, researcher, and industry analyst for over 20 years. Interviewed, cited, and quoted by media, think tanks, and academia, he has written for SecurityWeek, CSO Online and Dark Reading. While working at Gartner, he co-named the Security Orchestration, Automation and Response (SOAR) market, worked on the SIEM Magic Quadrant, and also covered the European MSSP Market. In past lives, Oliver worked for Qualys, Verizon, Gartner,  Tenable and Securonix and is currently Chief Furitist at Tenzir, where he works on product strategy and marketing.

Richard Cassidy

Richard Cassidy has been consulting to businesses on cyber security strategies and programs for more than two decades, working across...