Coalition Urges OpenAI to Scrap AI Ballot Measure Over Child Safety Concerns

A coalition of advocacy teams is urging ChatGPT developer OpenAI to withdraw a California poll initiative that critics say might weaken protections for kids and restrict authorized accountability for AI firms.

In a letter despatched to OpenAI on Wednesday, reviewed by Decrypt, the group argues that the measure would lock in slender child-safety protections, restrict households’ capability to sue, and prohibit California’s capability to strengthen AI legal guidelines sooner or later.

The letter, signed by greater than two dozen organizations together with AI coverage non-profit Encode AI, the Heart for Humane Know-how, and the Digital Privateness Info Heart, asks OpenAI to dissolve its poll committee and step again from the proposal whereas lawmakers work on laws.

“The principle demand right here is for OpenAI to withdraw from the poll,” Adam Billen, co-executive director of Encode AI, instructed Decrypt.

The dispute facilities on a proposed “Dad and mom & Children Protected AI Act,” a California poll initiative backed by OpenAI and Frequent Sense Media that will set up guidelines for a way AI chatbots work together with minors, together with security necessities and compliance requirements.

Within the letter, the teams argue that these guidelines fall quick. They are saying the measure defines hurt too narrowly, limits enforcement, and restricts households’ capability to convey claims when kids are harmed.

However OpenAI controls the precise poll initiative, Billen mentioned.

“OpenAI has the facility to withdraw it or put the cash in for signatures. All the authorized authority rests of their palms,” he mentioned. “They haven’t truly withdrawn the initiative from the poll. It is a frequent tactic in California, the place you place an initiative up and put cash within the committee.”

The letter factors to the initiative’s definition of “extreme hurt,” which focuses on bodily harm tied to suicide or violence, excluding a spread of mental health impacts that researchers and households have raised as considerations.

It additionally highlights provisions that will bar dad and mom and youngsters from bringing claims beneath the initiative and restrict enforcement instruments obtainable to state and native officers.

One other concern facilities on how the proposal treats person information. The teams argue that its definition of encrypted person content material might make it more durable to entry chatbot conversations which have served as key proof in latest lawsuits.

“We learn that as an try to dam households from having the ability to disclose their lifeless kids’s chat logs in courtroom,” Billen mentioned.

The letter additionally warns that the measure may very well be troublesome to revise if handed. It could require a two-thirds vote within the legislature to amend and tie future modifications to requirements equivalent to supporting “financial progress,” which advocates say might restrict lawmakers’ capability to answer new dangers.

Billen mentioned the initiative stays a think about ongoing negotiations in Sacramento, whilst OpenAI has paused its efforts to qualify it for the poll.

“They’ve $10 million within the committee, and you then say to the legislature, in case you do not do what we would like, we’ll put the cash in and get the signatures and put this on the poll, and if it passes, it should override regardless of the legislature does,” he mentioned. “So primarily, what’s taking place now could be they’re attempting to steer and management what state legislators do via using the initiative as a menace they’re leaving on the desk.”

OpenAI just isn’t the one firm going through scrutiny over chatbot-related harms. Earlier this month, the household of Jonathan Gavalas sued Google, claiming that Gemini pushed a delusion that escalated to violence and his final suicide. Billen, nevertheless, mentioned OpenAI’s method displays a broader sample within the tech business.

“The lobbying playbook that’s getting used on AI from these large guys specifically—the Googles, the Metas, Amazons—is identical technique that was used beforehand on different tech points,” he mentioned.

For now, the coalition is targeted on getting OpenAI to withdraw the measure and permit lawmakers to maneuver ahead via the legislative course of.

“It’s actually necessary, notably for the businesses which might be placing that expertise on the market, to not be those who’re writing the foundations that regulate them, as a result of that’s not significant protections,” Billen mentioned.

OpenAI didn’t instantly reply to Decrypt’s request for remark.