US Government Holds CEOs Accountable for AI Systems’ Unexpected Behaviour

The incident involving OpenAI and Hugging Face has shaken the AI world. While companies are now planning new safety measures and policies to advance AI development, the event has also raised a broader question for the sector: who should be held responsible when AI systems behave in unforeseen ways? Indeed, the US government is pointing to these companies, asserting that ultimate responsibility must lie with the people leading them-such as Sam Altman-rather than with the AI agents themselves.

Speaking to a famous publication on Monday, US Treasury Secretary Scott Bessent specifically addressed the Hugging Face incident, stating that OpenAI's leadership should be held accountable for what happened, rather than the AI agents that executed the actions. Bessent agreed with Daniel Huttenlocher, co-director of an MIT research lab, that "humans are responsible, not AI."

"The Hugging Face incident is the responsibility of OpenAI's leadership, not a group of agents," Bessent declared. He added that AI labs must take responsibility for their systems and noted that they can slow down development if they choose to do so. "These labs need to take responsibility for themselves," he stated, adding that "they can slow the pace whenever they want." These comments from the Treasury Secretary come amidst concerns within the AI sector regarding safety, cybersecurity, and the rise of increasingly autonomous AI agents. The United States and China are also discussing AI safety, including the potential creation of a system to report serious incidents involving this technology. Bessent indicated that the discussions would focus on risks such as uncontrollable AI agents, non-state actors, and cyberattacks.

The US government is also debating whether companies should be held legally liable when their AI systems cause harm. Bessent has opposed granting AI companies special protection against legal liability, arguing that such liability should rest with the individuals and companies that create these systems. During a House Financial Services Committee hearing held on September 15, he stated: "What we must not do regarding safety is grant these labs an exemption from liability, which is precisely what they are requesting." He further noted: "The best way to ensure safety is for creators to be held accountable for what they build and generate."

For his part, Anthropic CEO Dario Amodei has called for measures to slow down certain areas of cutting-edge AI development and has proposed greater cooperation among companies in the sector regarding safety.

However, Bessent has maintained that companies do not need to wait for the government to order them to slow down. "These labs need to take responsibility for themselves. They can hit the brakes whenever they want," he stated. He also noted that if human beings were carrying out similar actions, there would be consequences. "If people were doing this, one would expect repercussions and legal action," he said. "That is exactly what I think we need to do."

ABOUT THE AUTHOR

Kahekashan

[email protected]

Kahekashan is a passionate technophile with a keen eye for cutting-edge gadgets, emerging technologies, and everything in the digital realm. Raised in a Defence family with strong values and a background in literature, she has consistently pursued excellence in every endeavour. Her last full-time assignment involved content writing with the Indian School of Business.

Read More