USDA Is Using AI, but Doesn’t Have the Required Controls to Manage Risks, Watchdog Finds

Why It Matters

Without robust AI safeguards, USDA’s critical data and decision‑making processes risk compromise, setting a precedent for other federal agencies grappling with rapid AI adoption. Strengthening governance now protects national food‑security analytics and public trust.

Key Takeaways

• USDA uses AI for crop forecasts, supply‑chain risk, permitting advice
• No generative‑AI policy; key cybersecurity controls remain unimplemented
• Most FY2024 AI projects lack formal Authority to Operate approval
• Shadow AI risk grows as employees self‑report AI use only annually
• OIG recommendations target policy updates, risk management, and governance

Pulse Analysis

The Agriculture Department’s push to embed artificial intelligence across its operations reflects a broader federal drive to harness data‑driven insights for food‑security planning. By applying AI to yield estimates and supply‑chain monitoring, USDA hopes to accelerate decision‑making and reduce manual bottlenecks. However, the recent OIG audit underscores a critical blind spot: the agency has prioritized deployment over the foundational risk‑management framework mandated by the Office of Management and Budget, leaving its AI systems exposed to cyber threats and compliance gaps.

Federal AI governance is still evolving, and USDA’s experience highlights the challenges of scaling technology without parallel policy development. The lack of a generative‑AI policy and the absence of formal Authority to Operate (ATO) approvals for most FY2024 projects create fertile ground for “shadow AI,” where employees use unvetted tools that bypass oversight. This not only jeopardizes data integrity but also raises civil‑rights and critical‑infrastructure concerns, especially when AI influences permitting outcomes that affect environmental and economic stakeholders.

The OIG’s recommendations call for immediate policy revisions, the establishment of minimum risk‑management practices, and a systematic inventory of AI assets. Implementing these measures will align USDA with emerging federal standards, mitigate reputational risk, and ensure that AI-driven insights remain reliable and secure. For industry partners and agribusinesses, a more disciplined AI framework at USDA promises clearer guidance, reduced uncertainty, and stronger collaboration opportunities as the agency modernizes its analytical capabilities.