Adversarial AI exposes decades of dormant security flaws

Frontier AI models are surfacing software vulnerabilities humans missed for nearly three decades, creating unprecedented enterprise risk. Cognizant's Vishal Salvi urges simplified architectures and policies for non-human identities, estimating 50 such identities per human, to contain breaches faster.