US lawmakers unveil rules to regulate discriminatory AI

American congressmen renewed efforts to hold tech companies accountable for bias in algorithms. They introduced an updated version of the bill requiring audits of AI systems in areas such as finance, healthcare, housing, and education.

Proud to be introducing this much-needed legislation with @SenBooker and @RepYvetteClarke to hold companies accountable for reliance on flawed systems that discriminate against Americans based on age, gender, and race. https://t.co/GKJt5FpMPI

— Ron Wyden (@RonWyden) February 3, 2022

According to one of the bill’s authors, Senator Ron Wyden, the bill should lift the veil on the secret algorithms that decide whether Americans can see a doctor, rent a home, or gain admission to a school.

“If someone won’t rent you a home because of the color of your skin, that is outright illegal discrimination. The use of faulty algorithms or software that leads to discrimination and bias is also wrong,” the press release said.

The bill would require companies to audit their AI systems and report the results to the Federal Trade Commission (FTC). This includes any decisions made by enterprises, among which:

• the cost, terms, and availability of education;
• employment;
• basic utilities;
• family planning;
• healthcare;
• housing provision;
• legal services;
• any other services, programs, or opportunities affecting consumers’ lives.

The bill stipulates that the FTC would publish high-profile cases related to algorithmic bias. The Commission would also create a publicly accessible database where consumers can view which decisions are being made by automated systems at a given company.

The bill was backed by a number of civil rights groups. Arisha Hatch, vice president of Color Of Change, expressed hope that Congress would pass the measure soon.

Wyden first introduced the algorithmic accountability bill in 2019, but it did not proceed to committee hearings in either the House of Representatives or the Senate.

In April 2021, Wyden proposed to legislatively ban facial-recognition systems such as Clearview AI.

In November, lawmakers introduced rules for regulating recommender algorithms.

In January 2022, lawmakers proposed legislatively to ban targeted digital advertising on platforms such as Facebook and Google.

Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!