MITRE, a not-for-profit organization, announced that it will use its new compute capabilities, powered by an NVIDIA DGX SuperPOD, to train three new artificial intelligence (AI) foundation models for applications to critical government missions. Named the MITRE Federal AI Sandbox, this supercomputer will help train and deploy advanced AI models in critical infrastructure cybersecurity, weather modeling, and sustainable social services.
“AI has immense potential to transform the government’s service to its citizens and address important challenges ranging from improving resilience of critical infrastructure to making Medicare sustainable,” said Charles Clancy, MITRE senior vice president and chief technology officer. “The Federal AI Sandbox will be a resource for federal agencies to enable AI solutions.”
The Federal AI Sandbox provides an exaFLOP of 8-bit AI compute performance, making it capable of training new, large-scale foundation models. This immense computational power supports the development of AI tools, including generative AI, multimodal perception systems, and reinforcement learning decision aids.
MITRE is looking to train new foundation models designed to:
- Help cybersecurity experts prevent, detect, and respond to security breaches in critical infrastructure by analyzing massive and complex data to help a security operations center proactively identify and respond to threats.
- Enhance high-resolution weather modeling with improved precision and speed to help communities receive more accurate local forecasts and better prepare for urgent weather hazards.
- Revolutionize government agencies’ ability to provide citizens with fair and timely access to benefits by transforming millions of pages of law, agency policy, and regulations into tools that help streamline government workflows.
Agencies can access Federal AI Sandbox capabilities through existing contracts with any of the six federally funded research and development centers MITRE operates.
Earlier this month, MITRE’s Center for Threat-Informed Defense (Center) collaborated with more than 15 companies to increase community knowledge of threats and defenses for AI-enabled systems, culminating with the launch of the AI Incident Sharing initiative. The AI incident sharing website and submission form are available online.