helpnetsecurity.com·20 november 2026

AI agent API security hasn't caught up with agents already running in production - Help Net Security

Image for the article: AI agent API security hasn't caught up with agents already running in production - Help Net Security
Image via helpnetsecurity.com

Postman surveyed 9,000 developers and found that 25% of organizations already use AI agents in production, but many lack proper security controls. Four in ten respondents haven't established credentials and identity management for agents, and three in ten use the same credentials for agents as for other API clients. Only a third of companies maintain an inventory of their agents or monitor their activities. API-first companies are better organized but still have issues with shared credentials. Outdated documentation and the risk of AI-powered attacks and rogue agents are top concerns for the next two years.