Palo Alto: Firms Struggle to Secure Cloud AI Agents

Palo Alto: Firms Struggle to Secure Cloud AI Agents
A Palo Alto Networks survey of 2,800 security executives across 10 countries finds AI agents are nearly universal in software development, with 99% of organizations using them, yet security controls lag badly. Some 40% of organizations cannot reconstruct what an AI agent did, 53% log agent activity under human identities, and only 34% scan AI-generated infrastructure code before production. It takes an average of 18.4 days to fix security issues, and 60% still require manual intervention to revoke compromised credentials. Despite these gaps, 97% of respondents said they are at least somewhat confident in their AI supply chain security.
Read the original article →