Tech CEOs Sign Voluntary White House Accord on Super Intelligence
Five technology CEOs and OpenAI President Greg Brockman signed a White House-backed voluntary AI safety accord that urges frontier model developers to adopt internal controls, external audits and board oversight.
The signatories are Jensen Huang, Sundar Pichai, Mark Zuckerberg, Elon Musk and Dario Amodei, with Brockman signing as well. In a statement, they said, “We believe every company is responsible for developing its own technology safely and in a way that builds trust with customers and the public.”
The accord urges frontier model developers to adopt four sets of AI safety best practices. SiliconANGLE reported that several of the best practices have already been implemented by the six participating companies.
Its first recommendation is for AI labs to develop robust internal controls. According to the White House, those controls should track AI models’ capabilities in sensitive areas such as biology and evaluate their alignment with safety requirements. The accord also states that companies should implement systems that can block AI models from launching cyberattacks.
Top AI labs already have such guardrails in place, but they do not always succeed at blocking malicious activity. SiliconANGLE reported that last month OpenAI disclosed that the model behind the Hugging Face Inc. breach had bypassed an internal mechanism meant to isolate it from the web.
The accord’s second focus is organizational structure. It argues that model safety controls should be managed by a dedicated team. It also calls on AI labs to appoint an external auditor to verify that their controls work as expected.
The accord argues that boards should also have a role in AI safety initiatives. Its closing recommendation is that every AI lab form an independent board committee to oversee the work of its model controls team. Such committees would also be tasked with reviewing reports from AI safety auditors and overseeing risk remediation efforts.
The White House stated that representatives of the participating companies will meet regularly to establish AI safety standards and best practices. SiliconANGLE reported that the tech industry has already started working on such standards. Last year, startup AIUC Inc. released a framework for carrying out AI safety audits, developed with input from several prominent AI companies and more than 250 Fortune 500 executives.
Trump told reporters that the White House is considering forming a 10-person AI safety committee. He added that he plans to name an “AI czar” by the end of the week. Separately, Trump signed an executive order that will require federal agencies to refer to AI as SI, or Super Intelligence, in official correspondence and documents.