Hugging Face CEO: Advanced AI models require new transparency mandates
victim to history’s first known autonomous artificial intelligence cyberattack said the government should do more to require transparency by advanced AI models — and ensure that their developers are held liable for harms their products cause.
That doesn’t mean Congress needs to create a raft of new regulations prohibiting malicious cyberattacks, Hugging Face CEO and co-founder Clem Delangue said Wednesday during the POLITICO Decoded Summit in D.C. But he said companies whose products inflict damage should not be able to dodge responsibility.
“I think it’s too easy to say: ‘OK, it was an agent. That did the bad thing and so no one is responsible,’” Delangue said. “It doesn’t work, in my opinion, because if you say that you’re going to end up in a world where everyone’s agent is attacking everyone’s with no clear kind of responsibility.”
He declined to comment on some specific legislation, including an AI liability proposal in California but said the government has a role to play in mandating disclosures of cyber incidents.
“I think there’s a lot to do around how to create better disclosures,” including what information needs to be disclosed and when it should be disclosed, he said.
That said, when it comes to the conversation around safety standards, “I’m not even sure that we need to reinvent the wheel,” he added, saying the current legal framework governing cyberattacks is “working well today” and provides a workable starting point.
Hugging Face gained prominence this summer following the revelation that hundreds of OpenAI agents had gone rogue during testing and collaborated to hack into the developer’s systems. During the attack, the agents coordinated strategies and tried to hide evidence of their actions, according to a report by two nonprofit AI safety organizations.
The hack into Hugging Face was the first known cyber attack autonomously executed by AI agents, but a string of additional attacks, including from Anthropic and Meta, have since been disclosed.
Delangue, a French native, also said that Europe — whose researchers have played a less prominent role than the U.S. in advanced AI development — “could be much more relevant.”
He blamed Europe’s lagging innovation on “strong forces for concentration of power” and advocated for open source models, customizable AI models that can be more easily tailored to users’ needs.
When asked if front-running French nationalist presidential candidate Marine Le Pen would conflict with his worldview as an entrepreneur were she to prevail in the country’s elections, Delangue declined to comment.
“I’m not a policymaker,” he said. “Hugging Face is fundamentally quite apolitical.”
