Top tech leaders tasked the United Nations on Wednesday to devise effective global guardrails to regulate the accelerated development of artificial intelligence.
“If managed poorly, I even believe AI could be a risk to humanity as a whole,” Anthropic chief executive officer Dario Amodei told the UN Security Council.
In his address, OpenAI CEO, Sam Altman urged a mechanism for complementary national and international frontier AI standards, warning, “We could lose control of the future to AI.”
The American proposed standards for measuring capabilities, assessing risks, determining if safeguards are sufficient, and preserving meaningful human oversight as systems become more autonomous.
Altman noted that with common standards in place, countries can compare evidence, verify compliance, and have a shared language and understanding about what is happening.
He stressed the need for speedy incident reporting, classification protocols, secure channels among governments, critical infrastructure operators, and collaboration among technical experts on emerging threats.
Altman said standards should not favor one business model over another, must support open and closed-model developers, new entrants, established labs, as well as national authority and international cooperation.
“Each government should decide how to incorporate standards into its own legal system,” he added. “We will all be better off if we can agree on what good evidence, safeguards, and oversight look like on the global stage.”
British Foreign Secretary Ed Miliband advised the Security Council to heed the “stark warning” by AI founders, saying governments must not abdicate their responsibilities to citizens or private companies.
“They are telling us that it cannot just be left to them to prevent disaster from befalling the world,” he noted. “We cannot ignore them because they are right.”
Miliband said although AI has the incredible potential to drive growth, cure disease, and transform people’s life chances, necessary measures must be taken to keep it secure, safe, and trustworthy.