California privacy regulator approves watered-down AI rules

California’s privacy agency on Thursday greenlit controversial new rules dictating how to regulate automated decisionmaking technology (ADMT) used by the private sector.

The new regulations have been controversial because the California Privacy Protection Agency (CPPA) overhauled them to be significantly weaker than the originally-proposed rules. 

ADMT refers to technologies which rely on artificial intelligence algorithms to make decisions.

The most recent version of the proposed rules were released in May and immediately came under fire from privacy advocates and consumer watchdogs who were incensed by key language changes which greatly expanded the circumstances under which businesses can deploy ADMT. 

The rule allowing people to opt out now only covers decisions where the technologies "replace or substantially replace human decision-making." Previously, the language had been broader, covering technologies which "substantially facilitate human decision-making."

The revision of the proposed rules followed an intense lobbying campaign by industry and intervention from California Gov. Gavin Newsom, who personally argued for less stringent regulations. 

“The agency caved to companies and the governor,” said Justin Kloczko of Consumer Watchdog. “These rules will basically just formalize humans no longer being in the loop with regard to significant decisions being made in education, jobs and health care.”

In an interview with Recorded Future News last month, CPPA head Tom Kemp called the version of the regulations which were approved in a unanimous vote Thursday a “happy medium.”

Kemp acknowledged that the CPPA had modified the rules to take into account concerns from Newsom and industry, but argued that the now adopted regulations will nonetheless be “the best and the most robust” rules governing the use of ADMT in the country.

The regulations now head to California’s Office of Administrative Law and are expected to be finalized within a month.