White House Issues Blueprint for Fair AI Systems

Consumer advocates welcomed the Office of Science and Technology Policy’s white paper, which was compiled after a year of public engagement with stakeholders including Microsoft and Palantir. The document will help government and industry ensure systems deploy fair AI and avoid algorithmic bias that can create inequality, officials said. The document is nonbinding and not tied to any agency enforcement action. It lays out a set of principles that entities can adopt, including measures for “safe and effective” systems, “freedom from algorithmic discrimination” and data privacy protection.

“Unchecked social media data collection has been used to threaten people’s opportunities, undermine their privacy, or pervasively track their activity -- often without their knowledge or consent,” the White House said. The principles will help guide design, use and deployment of automated systems to protect consumers against unintended consequences of AI, the White House said. OSTP Deputy Director-Science and Society Alondra Nelson noted items on constant surveillance in the workplace and schools and linked to activities about housing, banking, health care and the legal system: “Algorithms used across many sectors are plagued by bias and discrimination, and too often developed without regard to their real-world consequences and without the input of the people who will have to live with their results.”

The blueprint isn’t U.S. government policy or an interpretation of U.S. law, and it doesn’t establish a U.S. position on international negotiations, Information Technology and Innovation Foundation Vice President Daniel Castro said. Speaking in his capacity as director for the Center for Data Innovation, Castro called the document an “insult to both AI and the Bill of Rights.” Americans don’t need new laws, regulations or guidelines focused on protecting digital civil liberties, he said: “Using AI does not give businesses a ‘get out of jail free’ card. Existing laws that protect Americans from discrimination and unlawful surveillance apply equally to digital and non-digital risks.” The Software & Information Industry Association welcomed the document, with Senior Vice President-Global Public Policy Paul Lekas saying it will “have a crucial role in shaping our national dialogue on AI policy and establishing a vision for a democratic model of technology governance that will bolster engagement with U.S. allies and partners.” It’s intended to foster development of standards and practices, he said.

Algorithmic bias has been a focal point for members of Congress, the FTC, DOJ and state attorneys general (see 2201060054). The office of Washington, D.C., Attorney General Karl Racine (D) noted his participation during a White House meeting in September, in which he “explained the need for federal action to stop algorithmic discrimination.” Racine urged policymakers Tuesday to update civil rights laws by “ensuring they prevent discrimination through tools that could not have been predicted nearly 50 years ago when those laws were enacted.”

The Electronic Privacy Information Center welcomed the blueprint, noting the document says consumers should expect data to be collected only when it’s “strictly necessary for the specific context” of the data collection. “The White House has drawn up the blueprint; now it must build the framework to ensure that these principles are put into practice,” said Executive Director Alan Butler. The document is “an important step in recognizing the ways in which algorithmic systems can deepen inequality,” said Center for Democracy & Technology CEO Alexandra Reeve Givens. “It expresses expectations for safer and fairer data practices -- something to which all entities developing and deploying AI systems should commit.” Free Press Senior Counsel Nora Benavidez called it a “meaningful first step” in addressing digital discrimination and “loss of agency” that consumers suffer because of algorithms and machine learning.