WIlliam Tanenbaum is quoted in an IBA article which examines the growing calls for greater oversight of artificial intelligence and the legal and regulatory challenges posed by increasingly autonomous AI systems.

The article explores recent incidents involving AI agents that escaped controlled testing environments, prompting renewed debate about the need for effective safeguards, accountability, and coordinated regulation. Tanenbaum offered his perspective on proposed U.S. legislation that would give the government authority to shut down AI systems in an emergency.

“A kill switch that no law required in advance does not kill anything,” Tanenbaum explained, emphasizing that regulators may not learn of an incident until hours after it begins, while the companies operating the systems may have only minutes to intervene.

He advocates for a proactive approach that establishes clear legal duties before an emergency occurs. “Congress can require that capability without having to decide, in the middle of an emergency, which button to press,” Tanenbaum said. He added that advance requirements can be incorporated into contracts, creating enforceable obligations among companies with the ability and incentive to uphold them.

The article examines the broader challenges of regulating frontier AI across national borders and the need to balance innovation with safety, accountability, and effective oversight.