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 may 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 slim 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 Data 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 Secure AI Act,” a California poll initiative backed by OpenAI and Widespread Sense Media that will set up guidelines for the way AI chatbots work together with minors, together with security necessities and compliance requirements.

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

However OpenAI controls the precise poll initiative, Billen mentioned.

“OpenAI has the ability to withdraw it or put the cash in for signatures. The entire authorized authority rests of their arms,” he mentioned. “They haven’t really withdrawn the initiative from the poll. It is a widespread tactic in California, the place you set an initiative up and put cash within the committee.”

The letter factors to the initiative’s definition of “extreme hurt,” which focuses on bodily damage tied to suicide or violence, excluding a variety of psychological well being impacts that researchers and households have raised as issues.

It additionally highlights provisions that will bar mother and father and youngsters from bringing claims underneath the initiative and restrict enforcement instruments out there to state and native officers.

One other concern facilities on how the proposal treats person knowledge. The teams argue that its definition of encrypted person content material may 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 with the ability to disclose their lifeless youngsters’s chat logs in courtroom,” Billen mentioned.

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

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

“They’ve $10 million within the committee, and then you definitely say to the legislature, if you happen to do not do what we wish, 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 occurring now’s they’re making an attempt to steer and management what state legislators do by means of using the initiative as a menace they’re leaving on the desk.”

OpenAI shouldn’t be 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 huge guys specifically—the Googles, the Metas, Amazons—is similar technique that was used beforehand on different tech points,” he mentioned.

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

“It’s actually necessary, notably for the businesses which can be placing that know-how 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.