GAO witness at Senate hearing on accountability within Law Enforcement for AI applications

Senate Judiciary Subcommittee on Criminal Justice and Counterterrorism Member Alex Padilla and Government Accountability Office Science and Technology Director Karen Howard discuss a GAO report which revealed lack of training for federal law enforcement agents, including the FBI and DEA, in protecting civil liberties when using facial recognition software, particularly concerning due to its demonstrated bias against black and brown individuals, lacking information on steps taken to address this issue, the importance of testing and verifying the accuracy and demographic bias of facial recognition algorithms, proposing that algorithms used for federal law enforcement purposes should be tested by independent third parties like NIST for accountability, the possibility of errors in both human and algorithmic analyses, and the need for accountability measures even when using advanced tools.