The US General Services Administration has added a trio of LLM vendors to its AI fast track procurement programme, smoothing their potential adoption by Federal agencies.
However, whether government departments are up to the task of exploiting the models safely may be open to question, after the Government Accountability Office reprimanded a trio of agencies for cybersecurity and IT procurement shortcomings.
The GSA has added what it described as “leading American AI companies’ products—Anthropic’s Claude, Google’s Gemini, and OpenAI’s ChatGPT—to its Multiple Award Schedule (MAS).”
GSA Acting Administrator Michael Rigas said in a statement, “We’re leveraging the private sector’s innovation to transform every facet of government operations.” This ranged from “streamlining back-office processes to revolutionizing employee and citizen experiences and reimagining how we deliver mission-critical services.”
Adding the models to the MAS programme would facilitate “strong, widespread federal agency adoption and ensure easy access to AI tools that will improve their everyday workflows and processes,” the statement continued.
The GSA emphasised that the announcement supported America’s AI Action Plan and that models came from “leading American companies.
Federal Acquisition Service commissioner Josh Gruenbaum added “We’re focused on models that prioritize truthfulness, accuracy, transparency, and freedom from ideological bias, aligning with the Trump Administration’s policy that federally procured AI systems must prioritize truth and accuracy over ideological agendas.”
How ready the Federal Government is to fully exploit those models is another issued. It emerged this week that the US Government Accountability Office, Congress’ spending watchdog, has written to the CIOs of three agencies pointing out a range of shortcomings around cybersecurity and IT procurement.
The CIO of the Environmental Protection Agency was informed that the agency has “11 open recommendations that call for the attention of the CIO. Each of these recommendations relates to a GAO High-Risk area: (1) Ensuring the Cybersecurity of the Nation or (2) Improving IT Acquisitions and Management.” These included one “priority” the letter said – establishing a process for “conducting an organisation-wide cybersecurity risk assessment.”
The Department of Homeland Security’s CIO was reminded the agency had “43 open recommendations”, of which seven were “priorities”, and four of which related to IT systems. These included conducting a full inventory of its hardware, software, and configurations, to help with network capacity planning.
And the GSA itself has four outstanding issues – including that the agency ensure that any AI application meet the demands of President Trump’s 2020 executive order on AI, and or be retired.