Scott Kupor calls for independent AI safety checks as White House weighs oversight
The task force’s vice chair described a role for outside audits and government verification, while urging developers to address data-center power and water demands.
Loading page…
The task force’s vice chair described a role for outside audits and government verification, while urging developers to address data-center power and water demands.
Listen to this story
Kupor's proposal would keep initial safety testing with developers but add independent security audits and a government backstop able to check compliance and enforce existing rules. The Oct. 12 interview did not set out task-force recommendations, new federal requirements, or a specific audit regime; the force is still developing proposals aimed at pairing responsible deployment with U.S. competitiveness. For operators, his comments signal scrutiny around pre-release testing and data-center resource burdens, but not a settled regulatory program.
Kupor said developers should test systems before release and notify the government when problems arise.
Independent security auditors could report findings to company boards, while government officials retain the ability to verify compliance and pursue enforcement.
He cited existing product-liability and corporate-responsibility laws, not a new AI-specific legal requirement.
AI companies should not be the only judges of their products’ safety, White House task force vice chair Scott Kupor said in an October 12, 2026, CNBC interview. He called for government verification alongside industry testing, while arguing against broadly slowing development. His comments outlined an oversight approach, not specific new recommendations or federal requirements.
Kupor, who also directs the Office of Personnel Management, appeared on CNBC’s Squawk Box to discuss the Super Intelligence Task Force’s work. The group is developing recommendations intended to preserve U.S. competitiveness while ensuring that increasingly advanced AI systems are developed and deployed responsibly.
President Donald Trump announced the force’s creation on October 4.
Kupor said developers should test products before release and notify the government when problems arise. He placed the initial responsibility on industry: companies must make sure the systems they release are safe and properly vetted.
One possibility he discussed was third-party security audits. Independent auditors could send their findings to corporate boards, while government officials would retain the ability to verify compliance and pursue enforcement.
For accountability, Kupor pointed to existing product-liability and corporate-responsibility laws. His remarks described those laws as part of the framework for holding developers responsible, rather than identifying a new AI-specific legal requirement. He also said safeguards would continue to evolve as systems become more capable.
Asked about disagreements among technology executives over AI safety, Kupor declined to speculate about their motives. He instead focused on developers’ conduct: companies that believe they are releasing systems they cannot control, and that could cause harm, should take their responsibilities under existing laws seriously.
We’re not just going to sit around and wait for the companies to tell us they’ve done their work.
Scott Kupor, speaking on CNBC’s Squawk Box, as quoted by MeriTalk
Kupor’s case for accountability did not extend to supporting a broad slowdown. He argued that restraining U.S. development would not stop competing nations from advancing their own capabilities.
Kupor also urged technology companies to address data centers’ electricity and water demands rather than wait for technological advances to ease those pressures. He suggested that companies could respond to concerns about rising electricity rates or provide their own power.
Those comments follow the administration’s voluntary Ratepayer Protection Pledge. Seven major technology companies signed it on March 4, committing to cover power-generation and infrastructure costs rather than pass them to consumers. The initiative expanded on July 23 to include utilities, state officials and data-center developers, but it imposes no binding federal requirements.
Kupor did not specify which measures the task force might recommend for resource demands, or whether it would seek additional federal requirements for data-center operators.
Loading discussion...
Join the conversation
Explain where independent audits are enough, or where government scrutiny adds value.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.