MITRE announces AI sandbox for federal agency use
By the end of the year, an artificial intelligence sandbox built by MITRE is expected to be operational for federal agency use, according to a Tuesday announcement from the nonprofit.
MITRE, which operates federally funded R&D centers on behalf of agencies, announced during the AI Expo for National Competitiveness that it anticipates applying the sandbox to “national security, healthcare, transportation and climate” according to the press release. The sandbox will be powered by AI data center infrastructure from NVIDIA.
MITRE said it is offering federal entities access to the tool’s benefits through existing contracts with R&D centers that the organization operates with agencies, including the Department of Homeland Security, the National Institute of Standards and Technology, the Department of Defense, the Federal Aviation Administration and others.
Charles Clancy, senior vice president and chief technology officer at MITRE, referenced President Joe Biden’s October AI executive order that outlines a series of responsibilities and deadlines for agencies to use the technology and support its implementation. Agencies “often lack the computing environment necessary for implementation and prototyping,” Clancy said.
“Our new Federal AI Sandbox will help level the playing field, making the high-quality compute power needed to train and test custom AI solutions available to any agency,” he added.
The sandbox provides computing power that is able to train AI applications such as large language models and other generative AI tools for government use, according to the press release.
MITRE said the supercomputer is also able to “train multimodal perception systems that can understand and process information from multiple types of data at once,” including images, environmental and medical sensors, text, radar and more.
Tuesday’s announcement comes nearly two months after the nonprofit announced the opening of a facility to test government use of AI for possible risks through red-teaming and human-in-the-loop experimentation.
