A push to have an international institution monitor artificial intelligence gained momentum as the U.N. General Assembly opened this week in New York City after several major technology leaders issued existential warnings about AI.
Norway and Finland this week led the declaration for “a call for control of frontier AI models,” which was adopted by 20 nations, most notably Germany and Canada, in addition to the European Commission. Among its recommendations was one urging the United Nations to explore creating “an international institution, able to set standards, enable verification, and convene states when capability thresholds are crossed,” which bears some resemblance to a global structure Anthropic CEO Dario Amodei recently pitched. Another recommendation urged governments and regional organizations to develop and coordinate “common standards” to ensure countries “across all regions have access to scientific capacity, expertise and trusted evaluation.”
“To realize AI’s potential, industry, governments and society must act now. We must address these risks and strengthen oversight — without widening the gap between countries in access to the benefits of AI,” the declaration says. “AI must remain under human direction, oversight and control. It must be developed and used in line with international law.”
The United States, China, the United Kingdom, France, Italy, Poland, and Japan are among the major global players that have declined to sign on to the declaration so far.
The development comes after two leading companies in the U.S., Anthropic and OpenAI, sparked the call for certain reforms and regulations for AI, arguing they are essential to keep AI from wiping out humans or circumventing human control of the future. Those warnings have been adopted by governments such as those that signed on to the latest declaration, but have been dismissed by critics as a push for more government control over technology, an opportunistic campaign by certain industry leaders to create a type of technology cartel, or a sinister plot designed to stoke fear of AI, leading the U.S. to lose the AI race to China and cede its dominant superpower status to Beijing.
In some aspects, the declaration mimics what Amodei called for in his recent letter for AI reform. That call has been adopted as essential for AI safety by some, including Sam Altman at OpenAI, while criticized by others as a subtle power grab disguised as a pitch to protect humanity.
ANTHROPIC CEO PITCHES AI SLOW-DOWN PLAN AFTER OPENAI-HUGGING FACE HACK: ‘WE OWE IT TO HUMANITY’
In his letter, Amodei suggested a new “global standards body” should be created that could mandate developers perform certain testing on AI models before release “for acute risks in areas such as cybersecurity, biology, and alignment.”
“I actually think creating such a body is likely feasible, but giving it real teeth will be a challenge, and the difficulty will be in verification that both sides don’t have secret models which they don’t test but may deploy in secret (e.g., for military applications),” Amodei wrote.