# Federal Agencies Seek AI Agent Controls

At the Oct. 21, 2026, FedInsider Innovation Summit, federal agency officials and industry representatives called for tightly scoped permissions, monitoring, external security controls, and ways to interrupt AI-agent activity. Participants differed on whether agencies can maintain human oversight at every step, highlighting a practical acquisition need for controls that limit agent access, enforce policy, provide visibility, and enable human escalation despite staffing constraints. The discussion signals a procurement priority, but no specific solicitation or contract award was announced.

- Agencies and contractors evaluating AI-agent solutions should assess least-privilege access, monitoring, policy enforcement, and interruption capabilities against their operational and security needs.
- Procurement teams should account for staffing constraints when defining oversight requirements, including when human review or escalation is necessary.
- The discussion points to demand for security controls that govern AI agents externally, rather than relying on agents to protect themselves.

**Jurisdictions:** federal
**Industries:** Information Technology
**Topics:** Artificial Intelligence, Cybersecurity
**Published:** October 23, 2026

### Government Entities
- U.S. Department of Agriculture (USDA)
- U.S. Customs and Border Protection (CBP)

### Vendors
- Fortinet Federal ()
- Versa Networks ()
- Dell Federal ()

### Key Quotes
> [The] human has always to be in the loop, and that’s the approach we’re taking.
> — Ajay Phogat, senior advisor to the chief information officer at CBP

> You can’t expect the model to protect itself.
> — Felipe Fernandez, chief technology officer at Fortinet Federal

### Sources
- [Agencies Urge AI Agent Access Limits, Weigh Human Oversight – MeriTalk](https://www.meritalk.com/articles/agencies-urge-ai-agent-access-limits-weigh-human-oversight) - MeriTalk