The discussions followed hiccups by AI companies who are facing growing pressure to establish common safeguards for increasingly capable models. According to CNN, who quoted close sources, the talks came on the heels of a July proposal by co-founder and chief executive officer of Google DeepMind, Demis Hassabis, for a US-led frontier AI standards body. According to the publication, in his July 14 proposal titled ‘Framework for Frontier AI and the Dawning of a New Age,’ Hassabis proposed a body modelled on the Financial Industry Regulatory Authority (FINRA), with independent technical experts and open-source representatives on its board.
He said the body should develop assessment protocols for frontier AI models and work with relevant federal agencies and US national laboratories on testing. Hassabis proposed that frontier AI companies initially submit models voluntarily for review up to 30 days before release. The assessments, he said, should cover capabilities in areas including cybersecurity and biological threats, while the framework could eventually be formalised as a requirement for frontier models deployed in the US. OpenAI’s chief scientist,
Jakub Pachocki had separately called for common safety requirements for AI development. In an essay published on September 6, Pachocki said existing commitments such as OpenAI’s ‘Preparedness Framework and Responsible Scaling Policy’ should evolve into widely mandated safety requirements. “We need to evolve commitments like the Preparedness Framework or Responsible Scaling Policy into widely mandated safety bars for continued development,” Pachocki said. He said the requirements could be enforced by third-party auditors, government agencies or international bodies.
At a briefing earlier this month, Pachocki confirmed that OpenAI was discussing possible standards with external organisations. “We’re talking to some external organisations about potential concrete standards we could put in place,” Pachocki added. He said the company would have more to share “in the next months”. OpenAI chief executive officer, Sam Altman in a post on X on Sunday said the company welcomed a federal framework establishing consistent safety requirements for frontier AI.
Altman said companies should not wait for legislation to begin developing common safeguards and backed shared standards for misalignment, monitoring and safety. “We hope that other companies will learn from our approaches and propose their own; we think shared standards for misalignment, monitoring, and safety will lead to better outcomes. We look forward to collaborating with our colleagues across the industry to formulate the best version of these,” the OpenAI boss said. He added that government action would be needed for international coordination, but that AI companies should begin doing what they can to establish responsible development practices.
