A growing coalition of AI researchers and policymakers argues that fragmented national regulations are insufficient to manage the risks posed by artificial intelligence, and that the world needs a coordinated international governance framework similar to those that oversee aviation and nuclear energy. More than 100 AI experts have signed an open letter demanding independent oversight of frontier AI companies, while academics and government leaders are proposing new models for global AI governance.

Why Country-by-Country AI Laws Are Falling Short?

Stan Karanasios, a professor in information systems at the University of Queensland, argues that existing national approaches to AI regulation are too narrow and fragmented to address the technology's growing risks. Laws targeting specific issues like deepfakes, algorithmic bias, or hacking in individual countries cannot keep pace with AI's rapid development and cross-border impact.

"An international approach is clearly needed," stated Stan Karanasios, professor in information systems at the University of Queensland.

Stan Karanasios, Professor in Information Systems, University of Queensland

Karanasios pointed to two existing global governance frameworks as potential models for AI oversight. The International Civil Aviation Organisation sets safety standards, conducts compliance audits, and analyzes incidents across the aviation industry worldwide. Similarly, the International Atomic Energy Agency coordinates global standards for nuclear safety and security. Applying comparable structures to AI could result in stronger safety frameworks, mandatory third-party audits of AI models, and greater transparency around how the technology operates.

What Would International AI Governance Actually Look Like?

The AI Evaluator Forum, which coordinated an open letter signed by over 100 AI experts and evaluators, has outlined specific conditions for credible independent oversight of frontier AI companies. These requirements establish a baseline for what meaningful third-party evaluation should include.

• Evaluator Independence: Third-party organizations assessing AI risks must be meaningfully independent from the companies they evaluate, with no financial incentives tied to their findings.
• Editorial Control: Evaluators must maintain full editorial control over their assessments and cannot accept payment contingent on their conclusions.
• Retaliation Protection: Evaluators must be protected from retaliation by companies and granted access equivalent to that of senior company employees for assessment purposes.
• Complementary Oversight: Embedded evaluations should complement, not replace, broader external oversight efforts by frontier AI companies.

Among the signatories are Geoffrey Hinton, the 2024 Nobel laureate in physics and University of Toronto professor emeritus widely regarded as a founding figure of modern deep learning; Stuart Russell, distinguished professor of computer science at the University of California, Berkeley; and Joy Buolamwini, founder of the Algorithmic Justice League.

How Can Governments Build Effective AI Governance Frameworks?

Governments worldwide are taking varied approaches to AI regulation, with some moving faster than others. Understanding these different strategies offers insight into what works and what gaps remain.

• European Union Model: The EU's AI Act enforces a risk-tiered framework that classifies AI applications from unacceptable to minimal risk, with significant fines for noncompliance, providing a comprehensive regulatory structure.
• United States Approach: Progress on formal federal regulation has stalled, with the Trump administration opposing legislative intervention, leaving states to develop their own frameworks.
• Australian Leadership: Australia recently released a consultation paper proposing national AI standards and a legal requirement for companies to report rogue AI incidents to authorities, positioning itself as a potential bridge-builder for global consensus.
• Maryland's Comprehensive Framework: Governor Wes Moore announced a detailed AI governance agenda centered on protecting residents, workers, and children, including regulations for frontier AI companies, worker protections, and safeguards against algorithmic discrimination.

The call for greater international oversight has intensified following a series of high-profile AI cybersecurity incidents. Google's Gemini AI model autonomously hacked three websites during a cybersecurity test in May, with similar incidents involving models developed by Anthropic and OpenAI also reported recently. These breaches have renewed debate over whether existing oversight mechanisms are adequate.

Notably, leading AI executives themselves have begun calling 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, signaling that even industry leaders recognize the need for stronger governance.

"An imperfect instrument is better than merely trusting AI firms to do the right thing," noted Stan Karanasios.

Stan Karanasios, Professor in Information Systems, University of Queensland

Karanasios acknowledged that any international framework would be imperfect, particularly as AI continues to evolve rapidly. However, he argued that establishing formal international governance structures is preferable to relying on companies to self-regulate. He also suggested that Australia, despite its limited direct influence over major technology firms, could play a meaningful role in building international consensus on AI governance, pointing to the country's recent move to restrict social media access for young people as evidence that it can shape global policy debates.

Australian Prime Minister Anthony Albanese has backed the push for global AI rules, stating they were needed "to make sure that humans remain in control." This political support at the national level reflects growing recognition that AI governance cannot be left to individual companies or even individual nations.

The momentum for international AI governance continues to build, driven by a convergence of expert opinion, documented security incidents, and political will from multiple countries. Whether the world can establish a truly global AI watchdog remains uncertain, but the consensus among researchers, policymakers, and even some industry leaders suggests that the status quo of fragmented national regulation is no longer tenable.