PRODUCTION SCENARIO
A retailer's gen AI program plans to launch a recommendation assistant in eight weeks and add security afterwards. The CISO asks for Google's recommended approach and a framework to structure the work.
What should the program lead propose?
Answering here is anonymous. Nothing is saved unless you sign in.
Show answer and explanation
Answer: Shift security into the design phase now, using the Secure AI Framework
Security and compliance controls are far cheaper to build into a design than to bolt on after development, which is why Google recommends shifting security left. The Secure AI Framework gives that work a structure across the whole AI lifecycle, including the attack vectors to assess such as data poisoning, model inversion, and adversarial attacks.