20 of the 35 AI offerings we track have a FedRAMP authorization on record: 11 at High, 6 at Moderate, 2 at Low, and 1 without a published level.
Authorization alone does not answer the CUI question. DFARS 252.204-7012 sets FedRAMP Moderate (or equivalent) as the baseline for cloud services handling covered defense information, so Low authorizations do not meet it.
For CUI, the ones with a "Yes, with conditions" answer are Claude for Government, Claude in Amazon Bedrock (AWS GovCloud), Claude on Vertex AI with Assured Workloads, Ask Sage, Kiro (AI coding IDE) in AWS GovCloud, Gemini for Government, Generative AI on Gemini Enterprise Agent Platform (formerly Vertex AI), Assured Workloads, Azure OpenAI in Azure Government, Microsoft 365 Copilot in GCC High, OpenAI models in Amazon Bedrock (AWS GovCloud), Grok models in Amazon Bedrock (AWS GovCloud), Amazon Q Business, Moveworks GovCloud, ChatGPT Enterprise and API Platform for FedRAMP, Codex (desktop app and CLI) signed in to a FedRAMP workspace and Zoom AI Companion for Zoom for Government.
Generated from our dataset. Sources last re-checked Oct 17, 2026.
Every row links to its primary source and full page. Statuses change; open the source before you decide. How we verify.
Related questions
What does FedRAMP authorized mean?
The offering is listed as FedRAMP Certified on the FedRAMP Marketplace, or sits inside a cloud provider boundary that is. The Marketplace now shows classes: Class B is Low, Class C is Moderate, Class D is High. 20x is the newer assessment process, not a level.
Which AI tools are FedRAMP Low only?
Adobe Acrobat AI Assistant (Document Cloud) (Low (LI-SaaS)) and Perplexity Enterprise and API Platform (Low).
Not legal advice. These answers describe what public authorization records and vendor documentation say, not whether your contract, data or configuration is compliant. Confirm with your contracting officer, assessor or counsel.