NTIA to Release AI Accountability Report 'This Winter'

The document will suggest policies for auditing AI systems, the agency's top official said.

NTIA to Release AI Accountability Report 'This Winter'
NTIA Administrator Alan Davidson at the Knight Foundation's Informed conference on Monday.

WASHINGTON, January 23, 2024 – The National Telecommunications and Information Administration is planning to release its artificial intelligence accountability report “this winter,” the agency’s top official said on Monday.

The NTIA first requested comment in April 2023 on what policies should be put in place to ensure the trustworthiness of AI systems.

More than 1,400 commenters weighed in, which is a lot for the agency, NTIA Administrator Alan Davidson said at a Knight Foundation event.

Davidson said audits will be a part of the policy ecosystem the report proposes.

“I think one of the things that we’ve seen is: like financial audits for the financial accounting system, there is going to be a role to play for audits in the AI ecosystem,” he said.

The agency’s report will weigh in on the data transparency requirements, safety standards, and consequences for violating those standards that will be necessary to make AI audits effective, according to Davidson. It will also consider how to stand up a market of third party AI auditors, another analogue to the accountability system that exists for financial institutions. 

In addition to assessing AI systems, the agency is also considering how policymakers can support trustworthiness “through the things that we fund” and “things that we press people to adopt,” Davidson said.

“Maybe there will be things that should be mandated or that we think policymakers or legislators should look at,” he continued.

The NTIA is set to formally request comment on another AI initiative “this spring,” Davidson said. That’s a study into how to regulate “open source” AI, systems whose model weights – key values that determine how the model functions – are publicly available.

The agency announced its intention to conduct the study in December 2023, a month after President Joe Biden released an executive order calling for inquiries into AI safety standards.

Popular Tags