This bill requires federal agencies that use, fund, procure, or oversee certain algorithms to create an office of civil rights focused on bias, discrimination, and other harms from those algorithms. A "covered algorithm" is defined as a computational process using machine learning, natural language processing, artificial intelligence techniques, or similar methods that can materially affect access to or terms of agency programs, regulated economic opportunities, or rights protected by an agency. The bill lists many "protected characteristics" (for example: race, color, ethnicity, national origin or immigration status, religion, sex and gender identity, disability, income level, age, veteran status, genetic information, and others).
Each civil rights office must employ experts and technologists on these issues. Each office must submit a report to relevant congressional committees within 1 year of the law taking effect and every 2 years after that. Reports must describe the state of covered algorithms in the agency’s area, risks of bias or discrimination tied to protected characteristics, steps the agency has taken to reduce harms, how the agency engaged stakeholders, and any recommended legislative or administrative changes. The bill also requires the Assistant Attorney General in charge of the Civil Rights Division at the Department of Justice to set up an interagency working group on covered algorithms and civil rights. The bill authorizes appropriations of such sums as may be necessary to carry out these requirements.
The bill states: "There are authorized to be appropriated to each covered agency such sums as may be necessary to carry out this Act." No specific funding levels or cost estimates are provided in the bill text or metadata.
No publicly available information.
No publicly available information.