MITRE Develops Secure AI Tool to Boost Productivity for US Government Agencies

"MITRE ChatGPT: Productivity for US Government Agencies" | CyberPro Magazine

(Source – CIO)

MITRE’s Pioneering AI Initiative

MITRE, a nonprofit research and development center for the US government, has launched a secure and advanced version of the popular generative AI tool, ChatGPT. Known as MITRE ChatGPT, this AI tool was developed internally and released in May 2023. 

MITRE ChatGPT is a customized version of Microsoft’s OpenAI GPT-4 and has already been adopted by more than 60% of MITRE’s 10,000 employees. The tool integrates with MITRE’s extensive 65-year-old knowledge base, enhancing productivity and efficiency across various projects.

The organization’s efforts have been recognized with a 2024 CIO 100 Award for IT leadership and innovation. MITRE ChatGPT supports six federally funded US agencies, including the Department of Defense, the Federal Aviation Administration (FAA), and the Department of Homeland Security. It enables swift analysis of legislation and policy documents, significantly reducing the time needed to deliver proposals to these critical agencies.

Transforming Operations with AI

MITRE’s Chief Information Officer, Deborah Youmans, and Director of Innovation and Experimentation, Michal Cenkl, are leading the charge to transform the organization into an “AI-native” entity. This transformation aims to leverage AI to provide intelligent and critical data to government agencies. Youmans, who became CIO in August 2023, emphasizes the importance of AI in MITRE’s future strategy, focusing on making the organization AI-native by integrating AI tools across its vast data network.

Since its release, MITRE ChatGPT has been continuously enhanced. By June 2024, it offered advanced document analysis and reasoning capabilities, provided an enterprise prompt library, and made the GPT-3.5 API available for various projects. The tool has also been upgraded to use GPT-4 and retrieval-augmented generation (RAG) for handling large documents. MITRE’s experience with AI systems like IBM’s Watson facilitated the rapid deployment of OpenAI’s GPT-4, enabling researchers to respond quickly to information requests from federal agencies.

Ensuring Security and Efficiency

Security has been a paramount concern in the development of MITRE ChatGPT. The AI tool operates within MITRE’s secure Microsoft Azure environment, with stringent security measures in place to protect sensitive data. MITRE negotiated with Microsoft to reduce logging and enhance security controls, ensuring that classified information remains secure.

MITRE ChatGPT has proven to be a versatile tool, assisting with coding, idea generation, problem-solving, and quality assurance. The tool has significantly reduced the time required to complete key tasks, with improvements ranging from 10% to 30%. To maximize its utility, MITRE hosts monthly AI workshops and technical exchange meetings to train employees on effective usage and to reinforce security guidelines.

Bill Hill, MITRE’s Chief Information Security Officer, has worked closely with the organization to ensure that MITRE ChatGPT adheres to strict security standards. This collaboration aligns with the Cybersecurity and Infrastructure Security Agency’s (CISA) Secure by Design principles, ensuring security is integrated from the ground up.

The demand for MITRE’s generative AI platform and robust AI hardware is expected to grow. According to Chirag Dekate, a vice president and AI analyst at Gartner, government agencies are increasingly looking to leverage generative AI to enhance productivity and streamline complex processes. MITRE’s AI innovations, including MITRE ChatGPT, are set to play a crucial role in this transformation, offering valuable tools for government agencies to build effective, value-centered solutions.

Also Read: Vulnerabilities in OpenAI’s ChatGPT Expose Users to Cyber Threats