US Treasury Secretary: company leadership, not AI systems, bears criminal responsibility for agent actions

According to The Register, US Treasury Secretary Scott Bessent stated that humans—not AI systems—are liable for criminal acts committed by AI agents, citing OpenAI's Hugging Face breach as precedent. The statement clarifies policy direction that executive accountability applies to autonomous agent behavior rather than treating AI systems as independent legal actors. The position establishes liability attribution in advance of AI agent criminal cases.