White House Issues Chief AI Officer Mandate for Federal Agencies

Federal agencies have 60 days to designate a chief AI officer who will be responsible for ensuring the government is minimizing AI-related impacts on civil rights and safety, Vice President Kamala Harris said Thursday. Harris announced OMB policies designed to protect against risks like bias and discrimination. The White House sent a memorandum to all executive departments and agencies, as directed under President Joe Biden’s AI executive order. It applies to civilian and military agencies, but there are exceptions and waivers for national security and law enforcement. The memo lays out safeguards, including impact assessments, that agencies must adopt when using AI. OMB said AI systems are “rights-impacting” if there’s a “legal, material, binding or similarly significant effect” on rights. In addition, agencies would have to apply specific transparency standards including the publication of AI uses and justifications. The memo grants travelers the right to opt out of airport facial recognition systems that the Transportation Security Administration controls. Exceptions and waivers for national security, intelligence and law enforcement could “significantly undercut” the document’s intentions, the American Civil Liberties Union said in a statement Thursday. Harmful and discriminatory uses of AI by national security agencies and state agencies “remain largely unchecked,” ACLU Senior Policy Counsel Cody Venzke said. The ALCU highlighted risks associated with law enforcement’s use of algorithmic systems like facial recognition and predictive policing systems, which produce “harmful results.” The Center for Democracy & Technology credited OMB for following stakeholder recommendations about improving agency transparency and government procurement of AI services. CDT said the White House missed an opportunity for establishing data minimization standards. Moreover, the administration should have provided a redress process if a chief AI officer “inappropriately grants a waiver,” CDT added. There’s “no recourse to challenge the validity of the decision to exempt AI uses.”