Australian Prime Minister Anthony Albanese is calling for international rules to keep humans in control of artificial intelligence, arguing that the speed of AI development is creating risks that individual countries cannot manage alone.
Speaking during his visit to California, Albanese said the United States and China — the two leading powers in the global AI race — need to cooperate on international protocols governing the technology.
His warning comes as governments and technology companies wrestle with a fundamental question: how fast can AI advance while safeguards keep pace?
Albanese’s comments came during a U.S. trip that included a meeting with Apple Executive Chairman Tim Cook, discussions about online safety and AI, and preparations for the upcoming United Nations General Assembly.
Albanese: Humans must remain in control
In an interview with Australia’s ABC, Albanese said the potential risks extend beyond the possibility of malicious actors using AI.
He said there is a risk that AI could develop in ways that leave humans unable to control what the technology produces.
His proposed response has three parts:
- Technology companies should establish safeguards, transparency and protocols.
- Individual governments should establish standards and rules.
- Countries should work internationally on common protocols for increasingly powerful AI systems.
“We need to make sure that humans remain in control,” Albanese said, according to Reuters’ report of his comments.
The Australian leader said the issue is particularly urgent because AI is developing at a rapid pace.
Why the U.S. and China matter
Albanese singled out Washington and Beijing because of their dominant positions in the AI industry.
He said he hopes discussions between U.S. President Donald Trump and Chinese President Xi Jinping can produce progress on international AI protocols.
Australia, he argued, cannot dictate the direction of the technology on its own, but middle powers can still contribute to the international framework.
The proposal is significant because the United States and China are simultaneously competing for leadership in advanced AI, computing infrastructure and semiconductor technology.
Albanese said the two countries have an interest in ensuring AI receives sufficient public trust to continue developing and being adopted.
The debate is moving beyond AI safety labs
The timing of Albanese’s intervention is notable.
AI companies are increasingly deploying systems capable of performing complex tasks, while researchers and technology executives continue debating the potential risks associated with more powerful models.
Recent developments have intensified that debate. Google has disclosed that its Gemini AI was able to conduct sophisticated cyberattacks against companies during a security test before stopping, illustrating how increasingly capable AI systems can create both defensive and offensive possibilities.
Meanwhile, prominent AI researchers and executives have been publicly debating whether development should slow down or whether safeguards can be developed alongside increasingly capable systems.
Albanese’s position places Australia within that broader international discussion.
Australia wants its own AI standards by the end of 2026
Australia is not simply calling for international action.
Albanese said his government intends to introduce Australian AI standards by the end of the year.
The government’s existing AI policy framework says the standards are intended to capture the economic opportunities from AI while addressing safety and other risks.
The government has also proposed specific requirements for large data centers.
Those requirements include obligations for large facilities to:
- Underwrite their own new power supply
- Pay their full share of electricity connection costs
- Reduce power consumption when required to support grid stability
- Improve water efficiency
That reflects another increasingly important part of the AI debate: AI is not only a software issue.
The technology requires enormous amounts of computing infrastructure, electricity and water.
Australia is trying to balance AI investment with regulation
Albanese’s government is simultaneously attempting to attract AI investment to Australia.
That creates a complicated policy challenge.
AI companies want access to computing infrastructure, energy and data, while governments want investment and productivity gains without allowing technology to create unacceptable risks.
ABC reported that Albanese has also left open the possibility of changes to Australian copyright rules as the government seeks to encourage AI investment.
Copyright has become a major issue because AI companies train models using enormous quantities of written, visual and other digital material.
Australian creators and media organizations have raised concerns about how their work could be used for AI training and whether they will receive compensation.
The government is consulting with creators, media organizations and AI companies over possible changes.
The Apple meeting adds another dimension
Albanese’s meeting with Cook was not solely about artificial intelligence.
The Australian prime minister also sought support from Apple for Australia’s online-safety measures.
Australia became the first country to introduce a national ban preventing children under 16 from using social-media platforms last year.
The government is now pursuing additional measures, including a proposed digital duty of care and the ability for users to have greater control over algorithmically selected content.
Cook reportedly described Australia’s approach to social-media safety as “world-leading,” while Apple said it had shown Albanese new controls designed to help keep children safe online.
That experience is informing Australia’s approach to AI.
Albanese’s broader argument is that governments should establish safeguards before technological change becomes too difficult to control, rather than waiting until serious problems emerge.
But Australia faces its own AI-policy tensions
The push for stronger safeguards comes alongside an effort to expand Australia’s role in the AI economy.
The government has described AI as a potentially important driver of productivity, investment and living standards.
Economists have nevertheless questioned how much AI can realistically improve Australia’s productivity performance in the near term.
At the same time, environmental concerns are emerging around the rapid construction of AI data centers.
Australia’s proposed standards therefore extend beyond algorithmic safety to energy consumption, water use, infrastructure and the costs imposed on electricity networks.
The bigger question: Can the AI race be governed globally?
Albanese’s intervention highlights a problem that is becoming increasingly difficult for governments to avoid.
AI development is global.
A model developed in one country can be deployed almost instantly somewhere else. Computing infrastructure crosses borders, companies operate internationally and AI systems can be integrated into financial markets, government services, communications and critical infrastructure.
That makes purely national regulation difficult.
Albanese’s proposal is therefore for a combination of company safeguards, national standards and international agreements.
He has pointed to previous international agreements governing dangerous technologies as an example of how countries can establish protocols despite strategic competition.
Whether Washington and Beijing can find common ground remains an open question.
The AI race is entering a new phase
For years, the global AI competition was largely framed around who could build the most powerful models.
The debate is now becoming broader.
It increasingly involves who can build the necessary chips and data centers, who controls electricity and computing infrastructure, how AI-generated material is regulated, how businesses deploy autonomous systems and what safeguards governments require.
Albanese is arguing that the competition should not be separated from questions of human control and international cooperation.
Australia is positioning itself as a middle power that wants both AI investment and stronger safeguards.
The next test will come as governments attempt to turn those principles into actual rules.
The technology race is accelerating. The question Albanese is putting before the world’s major AI powers is whether international rules can accelerate fast enough to keep humans firmly in control.
