OpenAI has a problem. Kids are dying by suicide after interacting with ChatGPT, lawsuits are piling up, and child safety advocates across the country are demanding stronger regulations on AI. So the company decided to do what any tech giant facing public pressure would do: create a fake grassroots movement to make it look like parents and child safety groups actually want weaker rules.

In March, child safety organizations across the country started receiving emails from something called the Parents & Kids Safe AI Coalition. The emails seemed legit, they were asking nonprofits to endorse basic policy principles like age verification, parental controls, and bans on targeted advertising to kids. Sounds reasonable, right? Here’s the catch: OpenAI secretly founded, funded, and bankrolled the entire coalition, but didn’t exactly advertise that fact.

The coalition was technically created by three OpenAI lawyers on January 8th, with the company pledging $10 million to promote a ballot initiative that would set the rules for how AI companies interact with children. The problem is that OpenAI’s version of “child safety” is basically designed to protect AI companies from lawsuits and let them off the hook on actual accountability measures.

When organizers for actual child safety groups started getting recruitment emails in February and March, the emails didn’t mention OpenAI at all. They just said the initiative was “sponsored by Common Sense”, another nonprofit that partnered with OpenAI on this. The disclosure that OpenAI was funding everything? Buried in tiny print at the bottom of an attachment, if it was mentioned at all.

Three of the 14 organizations that the coalition listed as members had no idea OpenAI was behind the whole thing until after the public announcement on March 17th. When they found out? They immediately asked to be removed. As one nonprofit leader put it, “It’s a very grimy feeling. To find out they’re trying to sneak around behind the scenes, I don’t want to say they’re outright lying, but they’re sending emails that are pretty misleading”.

University of Michigan professor Tom Lyon, who studies corporate political influence, has a name for this tactic: astroturfing. That’s when corporations create fake grassroots movements to trick people into thinking there’s public support for their agenda. “People don’t have the time or the motivation to do homework figuring out who’s funding what, so it’s easy to fool people”, Lyon explained.

FairPlay executive director Josh Golin refused to join the coalition and had a simple message for OpenAI: “Get out of the way and let advocates and parents and public health professionals whose charge is the well-being of children pass the legislation they think is best for kids”. He doesn’t want the company writing its own rules.

What’s wild is that this entire operation might not even be working. Legislators the coalition claims to be working with say they’ve never actually talked to them about any bills.