# Contractors Address Agentic AI Fraud Risks

An article for government contractors examines how existing procurement-fraud laws could apply when autonomous AI agents take improper actions in government contracting. Potential legal exposure may depend on the agent’s interaction with government systems—particularly whether it submits information or only accesses a system without authorization—and on the intent, direction, or knowledge attributable to people and the contractor. Contractors using agentic AI for proposal-related work should establish human review, authorization limits, and audit trails; the signal describes legal analysis, not a change to procurement law or a specific enforcement action.

- Companies using AI agents in proposal preparation or submission should define which actions require human approval and limit agent permissions for government-system access and submissions.
- Maintain audit trails that document agent activity and relevant human direction or review, since attribution of intent and knowledge may affect legal exposure.
- Procurement and legal teams should assess how their agentic AI tools interact with government systems, distinguishing submission activity from unauthorized access when evaluating controls and risk.

**Jurisdictions:** federal
**Industries:** Information Technology
**Topics:** Artificial Intelligence, Regulatory Compliance
**Published:** October 10, 2026

### Sources
- [Depending on the interaction with the government system, there are already much stronger legal hooks - especially if it involves submissions rather than just unauthorized access. I wrote this with govt contractors in mind, but its application is broader. https://t.co/Z2iYM3FCYW](https://x.com/JTillipman/status/2108955082255741154) - twitter-fed-procurement