AFR: What Directors Need to Know and Do About Frontier AI Risks

The Australian Financial Review's Agnes King, in its cybersecurity report published on 30 July 2026, walked Australian directors through the rare Five Eyes agencies alert on frontier AI models capable of complex problem-solving, multi-step planning and autonomous execution across text, code and images. Kathryn Giudes was included as an expert voice alongside Palo Alto Networks' Nicole Quinn, Mallesons cyber practice head Cheng Lim, the AICD's Louise Petschler and veteran director Pat Sullivan.

Kat's contribution framed the speed and cost collapse of AI-driven attacks: what once took months of preparation can now be executed in minutes, at trivial cost, by someone with no computing background. The piece cited the CodeWall hack of McKinsey's internal AI platform earlier this year, achieved for USD $20 in two hours, as the prototype of what boards are now being asked to defend against. Kat also flagged the sovereignty exposure of AI stacks built on US-hosted models subject to the CLOUD Act.

"For the cost of a cup of coffee someone can cost a company millions."

Kathryn Giudes, Founder, ORCA Opti (Australian Financial Review, 30 July 2026)

Read the full feature at the AFR: What directors need to know and do about frontier AI risks (Australian Financial Review, by Agnes King)