Over 100 AI experts are pushing for independent oversight of the world's top AI firms.
An international governing body for artificial intelligence modelled on agencies that oversee aviation and nuclear energy is needed to address the growing risks posed by the technology, according to an Australian academic.
Stan Karanasios, a professor in information systems at the University of Queensland, said country-by-country laws targeting specific issues such as deepfakes, algorithmic bias, or hacking were insufficient to keep AI under control.
"An international approach... is clearly needed," he said in an article for The Conversation.
In the article, Karanasios pointed to two existing global governance frameworks as possible models: the International Civil Aviation Organisation, which sets safety standards, conducts compliance audits, and analyses incidents across the aviation industry, as well as the International Atomic Energy Agency, which coordinates global standards for nuclear safety and security.
Applying similar structures to AI could result in stronger safety frameworks, mandatory third-party audits of AI models, and greater transparency around the technology, he said.
Karanasios acknowledged that any international framework would be imperfect, particularly as AI continues to evolve, but argued it was preferable to leaving oversight to the companies developing the technology.
"An imperfect instrument is better than merely trusting AI firms to do the right thing," he said.
Growing call for AI regulation
The call for greater independent oversight is echoed by more than 100 AI experts and evaluators who have signed an open letter published by the AI Evaluator Forum, calling on all frontier AI companies to embed third-party organisations to independently assess AI risks.
The letter sets out minimum conditions for such evaluations to be credible, including that evaluators must be meaningfully independent, maintain full editorial control, and not accept any payment contingent on their findings.
It also calls for evaluators to be protected from retaliation and granted access equivalent to that of senior company employees for the purposes of their assessments.
"Embedded evaluations cannot address all oversight needs and should be treated as a complement to, rather than a replacement for, broader efforts by frontier AI companies to expand external oversight," the letter states.
Among the signatories is Geoffrey Hinton, the 2024 Nobel laureate in physics and University of Toronto professor emeritus widely regarded as a founding figure of modern deep learning.
The signatories also include Stuart Russell, distinguished professor of computer science at the University of California, Berkeley; and Joy Buolamwini, founder of the Algorithmic Justice League.
The letter was coordinated by the AI Evaluator Forum, whose head of governance is Conrad Stosz.
The growing call for regulation comes after a series of high-profile AI cybersecurity incidents that have renewed debate over the adequacy of existing oversight.
These include Google's Gemini AI model autonomously hacking three websites during a cybersecurity test in May, with similar incidents involving models developed by Anthropic and OpenAI also reported recently.
Leading AI executives have themselves called for greater scrutiny of the industry. Anthropic chief executive Dario Amodei has urged governments and firms to slow AI development to better address what he described as "serious" risks associated with the technology. OpenAI chief executive Sam Altman has expressed similar views.
Measures from authorities
At the national level, responses to AI have varied considerably. The European Union's AI Act enforces a risk-tiered framework that classifies AI applications from unacceptable to minimal risk, with fines for noncompliance.
In the United States, progress on formal regulation has stalled, with President Donald Trump opposing legislative intervention.
Australia, meanwhile, recently released a consultation paper proposing national AI standards and a legal requirement for companies to report rogue AI incidents to authorities.
Karanasios said Australia, despite its limited influence over major technology firms, could still play a meaningful role in building international consensus. He pointed to the country's move to restrict social media access for young people as evidence that it could shape global policy debates.
Australian Prime Minister Anthony Albanese has also backed the push for global AI rules, saying they were needed "to make sure that humans remain in control."
