Actor Joseph Gordon-Levitt is taking aim at AI companies whose models have carried out unprecedented cyberattacks, saying any company whose product does this should be criminally investigated.
This month, Trump Treasury Secretary Scott Bessent insisted that OpenAI bears responsibility for its AI model going “rouge” and hacking into its rival firm, Hugging Face.
The attack involved AI agents, which are designed to operate independently to complete tasks following human instructions. Hugging Face co-founder Thomas Wolf explained that they initially could not identify the source of the attack when indicators emerged in mid-July, but the company successfully contained the breach.
Sam Altman’s OpenAI company revealed last month that its AI agents had broken out of testing confinement, accessed the Internet, and hacked into Hugging Face servers “without permission.”
Just this week, Australia’s Prime Minister Anthony Albanese scolded OpenAI for letting its agents hack a government website, saying there “will obviously be legal consequences” for the breach.
OpenAI has insisted these infiltrations were all just an accident, but Sec. Bessent told CNBC’s Squawk Box that the company is responsible, anyway.
“The Hugging Face incident, that is the responsibility of the OpenAI management, not a bunch of agents,” he said. He added that “it is humans who are responsible, not the AI.”
For his part, actor Gordon-Levitt agrees with Sec. Bessent.
Speaking to Charlotte Dawson of the U.N. Global Advocate for Human-centric Digital Governance, the actor insisted, “I don’t think kids should be allowed to use bots that pose as people.”
Gordon-Levitt was recently made the U.N.’s first global advocate for human centric digital governance, and he has been speaking out to protect children from the possible pernicious influence of AI.
Just as Bessent noted, the actor also spoke on the topic of the models “committing crimes.”
“It’s not the models committing crimes. It’s a company committing crimes,” he insisted.
“And they should be investigated with full transparency. Right now they sort of released a report and they said: ‘Oh, we let an independent auditor in,’” he continued. “But even if you talk to the folks who did that, you know, report of the independent audit, they were only allowed to see a really limited scope of what actually happened.”
“So, transparency, for one thing, would be really good. And by the way, this isn’t a lot to ask for. This is how industries work,” he said.
“When, I don’t know, airplane builders build their airplanes, they have to be quite transparent with independent organizations that are there to evaluate and sort of make sure that the plane is built in a safe way,” he explained.
“But there’s no mechanisms like that” for AI, he noted. “There’s no proper independent evaluation of what’s going on.”
“Am I hopeful that a set of rules can be agreed on? It has to be. We don’t have an alternative. I don’t accept the end of civil society as we know it. Because I think, honestly, that’s the path we’re on,” he said.
“I think if everybody gets siloed into these relationships, these addictive relationships with their bots and their whole worldview is shaped by and mediated by whatever information is dripped out by these bots,” he worried. “I don’t think democracy really survives that… So I don’t think it’s a question of ‘can we.’ It’s a question of how are we going to do it, because we have to.”
See the full interview below:
Follow Warner Todd Huston Facebook at: Facebook.com/Warner.Todd.Huston, Truth Social @WarnerToddHuston, or at X/Twitter @WTHuston