Escape AI Quicksand: Data Misuse, Misadventures, And Mal-Intent
CxOTALK, Friday, October 3rd, 2025
As you rush to deploy generative AI in your organization, are you stepping into treacherous terrain where bad data, poisoned inputs, and broken guardrails threaten to undermine even your most sophisticated AI initiatives?
In episode 896, host Michael Krigsman talks with Dr. David Bray and Dr. Anthony Scriffignano to expose the hidden dangers lurking beneath AI's promising surface and reveal strategies for building trustworthy, resilient AI systems.
Your AI landscape faces two fundamental challenges that can create wrong, even catastrophic, conclusions from your data:
Adversarial Attacks: Intentional manipulation through deepfakes, misinformation, disinformation, and deliberately poisoned datasets designed to corrupt your AI outputs and decision-making.
Naive Implementation: Errors arising from treating generative AI like autopilot, deploying it without proper calibration, guardrails, or understanding of its limitations, leading to misuse and misadventure.