California Attorney General Rob Bonta said his office is looking “very closely” at OpenAI and Anthropic’s businesss in the wake of the safety warnings issued by the AI companies themselves.
“It’s a top priority for our office,” Bonta told TheWrap CEO and editor-in-chief Sharon Waxman during a spotlight panel at our TheGrill business conference Wednesday. “This is all hands on deck, in my humble opinion.”
Bonta said his office was looking at the “Hugging Face” incident, in which OpenAI’s AI agents got out of its containment environment and attacked the AI platform. While no serious damage was done, it was further evidence of AI companies’ potential inability to control and predict what its systems could do.
Bonta’s comments also come in the wake of Anthropic CEO Dario Amodei’s warning that AI companies need to slow down the pace of their model development for the sake of safety, which followed social posts from former and current Anthropic employees about the existential danger AI poses to the world.
Bonta told TheWrap on the sidelines of the conference that President Trump’s pro-AI stance puts more of the onus on him and other states to put guardrails on the AI companies, an echo of a statement he made on the panel.
“Congress needs to ask here: a president who says that the threat of AI is a hoax–that’s not helpful,” he said. “Congress who’s not willing to pass laws that protect us from AI; that’s not helpful either.”
But not everyone in Congress is standing idly by. Sen. Bernie Sanders (I-Vermont) and Rep. Greg Casar (D-Texas) introduced the Ban Artificial Superintelligence Act, which would permanently ban the development or deployment of AI superintelligence, the state in which AI becomes smarter than humans. Rep. Ted Lieu (D-California) introduced the AI Kill Switch Act.
Bonta expressed support for the idea of implementing a kill switch. He also told TheWrap that he believed state enforcement could work hand-in-hand with congressional legislation to better add protections, and agreed that any action he would take could fuel the legislation process.
A representative for OpenAI declined to comment.
“The more watchdogs, the better. The more regulators, the better,” Bonta said. “Making sure that we’re safe from catastrophic risk and the end of humanity, and from kids being taught how to kill themselves and having sexually explicit conversations with chatbots and everything in between.”
