AI trustworthiness

Mitre Opens New Lab in Virginia to Assess AI Risks

Mitre has opened its new AI Assurance and Discovery Lab to evaluate potential risks associated with the federal government’s use of artificial intelligence, FedScoop reported.

Located in Virginia, the lab will assess the risks of systems using AI through simulations, red-teaming and “human-in-the-loop” testing. The facility will also test for bias in systems.

The lab’s creation comes amid concerns about the trustworthiness of AI and aligns with President Joe Biden’s executive order, which emphasized the need to mitigate the risks of AI in government.

During the lab’s opening event, Charles Clancy, Mitre’s chief technology officer, said the company envisions the lab to be a place where it can pilot and develop concepts of AI assurance.

Sen. Mark Warner and Reps. Gerry Connolly and Don Beyer also attended the event to express their support. Warner cautioned that the rapid development of large language models could outpace safety standards and underscored the urgency of establishing guidelines in the evolving AI landscape.

Mitre’s new lab will use its proprietary AI Assurance Process to guide comprehensive risk evaluations. Miles Thompson, Mitre’s AI assurance solutions lead, explained that the level of assurance involved in the process will vary depending on the system and its intended use.