NYC begins enforcing new law targeting bias in AI hiring tools

New York City’s Automated Employment Decision Tool (AEDT) Act, considered the first in the United States to reduce bias in recruitment and employment decisions based on IA, will now be enforced – after the law came into effect in January and the final rules were adopted in April.

Under the AEDT Act, it will be illegal for an employer or employment agency to use artificial intelligence and algorithm-based technologies to assess NYC applicants and employees – unless it conducts an independent bias audit before using AI employment tools. Bottom line: New York City employers will be the ones to take on the compliance obligations around these AI tools, rather than the software vendors who create them.

Technically speaking, the law took effect on January 1, but in practice companies could not easily comply because the law did not provide enough detail on how to comply with a biased audit. But now the city’s Consumer and Worker Protection Department has released an FAQ intended to provide more details.

Companies should conduct an annual AI bias audit

According to the FAQ, the bias audit must be performed annually, be “an unbiased assessment by an independent auditor” and, at a minimum, “include selection or scoring rate calculations and the category impact report. gender, race/ethnicity, and intersectional categories.

The law requires employers and employment agencies to comply with “all relevant anti-discrimination laws and rules in determining what action is necessary based on the results of a bias audit” and to publish a summary of the results of the most recent bias audit.

According to Niloy Ray, shareholder of the labor and employment law firm Littler, in the majority of cases, complying with the law should not be particularly difficult, but it does require collaboration between third-party providers who create tools hiring AI and the companies using them.

“The law contains a fairly dense description of the technologies to which it applies, which requires understanding how the tool works,” Roy said. “They’re going to have to explain it enough to help businesses [do the bias audit]so it’s a good result.

That said, there are extreme cases where it may be more difficult to determine whether the law applies. For example, what if the job is an entirely remote position? Does New York City have jurisdiction over this role?

“These edge cases get a little more confusing, but I think in general it’s still easy as long as you can understand the technology,” Roy said. “Then it’s just a matter of collecting the data and performing some simple arithmetic on the data.”

Roy pointed out that New York isn’t the only state or jurisdiction considering this type of law governing AI bias in hiring tools. “California, New Jersey, Vermont, Washington DC, Massachusetts, they all have versions of regulations running through the system,” he said.

Discuss bias audit requirements with vendors and legal counsel

But in New York, any large company that is hiring is likely ready with what it needs to comply, he added. For small businesses, the vendors they acquire tools from have likely already done this bias audit.

“If you are working with a tool that you did not develop but obtained from a third party, contact them immediately and discuss what they can do to help you be compliant,” a- he declared. “On the internal side, you may need to contact your legal counsel, someone who does this for several or hundreds of companies, and they will be able to give you a quick boost with a framework.”

Even for those who missed the July 5 deadline, it is important to continue working to ensure compliance is carried out as efficiently as possible and to document your efforts to obtain legal advice and assistance from suppliers.

“It makes a huge difference if you say I put my head in the sand rather than I saw the train coming, I couldn’t make it to the station, but I’m still trying” , explained Roy. “If you work in good faith, [they’re] won’t penalize you [they’re] is not going to sue, given the newness and complexity of the law.