Uncle Sam Speedruns AI Chatbot Adoption For Federal Workers

The US government wants more AI chatbots in fed employees' hands, and its push to do so means that tech companies keen to provide other services will have to get in line behind the LLM makers.

The General Services Administration is going to begin prioritizing authorization of cloud services that include "conversational AI engines designed for routine and repeated use," according to an announcement on Monday. AI products get to jump the approval line for FedRAMP, the US government's cloud software security certification program. 

Instead, AI companies will use GSA's FedRAMP 20x, a Biden-era program designed to streamline the government's cloud software approval process. FedRAMP 20x's goals include the automation of FedRAMP security verification by standardizing required documentation, initiating a rolling validation program that continuously monitors application changes, and deploying industry security standards. 

The GSA hopes to "streamline the adoption of advanced AI capabilities across the federal government" by accelerating approval of AI offerings that include chatbots under FedRAMP 20x, the agency said.

But while AI chatbot firms might get to skip the FedRAMP line, getting approval still won't be easy. According to the FedRAMP AI prioritization page, criteria to meet the bar for prioritization is steep, and one must meet each of five requirements:

1. Enterprise-grade security (SSO, real-time analytics, SCIM provisioning, etc.)
2. Guaranteed protection of data, including an agreement that training data won't leave the customer's environment
3. Established demand from at least five CFO Act agencies or be recommended by the CIO Council
4. Be "available for government use via" a Multiple Award Schedule (MAS) 
5. Be able to meet FedRAMP 20x standards within two months of acceptance 

FedRAMP 20x standards are tough enough, but the fourth of those five criteria might mean that, yet again, the big AI companies are getting a boost while smaller organizations are left behind. 

• Box's AI agents set to help US government agencies
• Uncle Sam asks industry if it has AI that'll make procurement suck less
• GSA inks another $1 OneGov vendor deal, this time with Anthropic
• Uncle Sam launches AI sandbox for government use, says it won't be around for long

Multiple award schedules are a GSA contract type that allows companies to sell products to government agencies at the state, federal, and local levels at prenegotiated terms. The goal is to streamline and speed the acquisition process for agencies; as such the Trump administration has made MASes a key part of its AI action plan.

OpenAI, Google, and Anthropic all signed MAS contracts with the GSA for AI services this month, meaning that they're all one step closer to getting approval prioritization for future federal government AI offerings. It's not clear from the language on the GSA AI prioritization page whether eligible companies have to already have an MAS or not. 

We reached out to the GSA for more info, but didn't immediately hear back. As of writing, no AI companies have demonstrated "their full commitment and ability" to meet the GSA's AI prioritization standards. ®