Australia's AI Plan Is Bold. The Tech Giants It Must Deal With Are Bigger Than Most Governments

Key points

• Australian Prime Minister Anthony Albanese outlined his government's AI policy in a speech at the University of Sydney in 2025.
• Albanese said Australia would aim to "get out in front" of artificial intelligence, not just react to it.
• The tech giants that set the rules for AI products operate across borders and can set their own terms with individual countries.
• Regulating social media has already shown how hard it is for national governments to hold these companies to account.
• The biggest AI companies have market valuations that dwarf Australia's entire federal budget.

Anthony Albanese stood at a lectern at the University of Sydney this week and promised something ambitious. His government, he said, wouldn't just keep pace with artificial intelligence, the technology that uses computers to perform tasks humans usually do. Australia would get out in front of it.

It's a striking phrase. The harder question is what it means in practice when the companies building that wave answer to no single government.

Look at what happened with social media. Platforms like Facebook and TikTok spent years making their own rules about what speech was allowed, what data they collected, and how much they charged governments for access to their systems. Countries that pushed back found the companies willing to restrict services or dig in for years of legal argument. Australia has been bolder than most: it banned children from social media platforms late last year. But the fight was long and expensive, and the outcome is still contested.

AI is the same problem, scaled up. The companies at the frontier of this technology, building the large language models (AI systems trained on vast text so they can read, write and reason), are American firms with cash reserves and lobbying budgets that dwarf what most governments spend on technology policy in a decade.

That's not a reason for Albanese to give up. It's a reason to be precise about what "getting out in front" actually means, as The Guardian reported after the speech. We covered Albanese's plan to create a single national office overseeing AI's economic, security and social risks on 14 July 2026, and policy experts were already warning that accountability guardrails need to come before permits do.

What does this mean for ordinary Australians?

For most people, it means the rules around how AI is used at work, in hospitals, in courtrooms and in schools will be shaped partly in Canberra and partly in San Francisco and Seattle. What Australia can realistically control is how AI is used inside its own institutions: which tools government agencies can buy, what data those tools can access, and what happens when an AI system harms someone.

That's a meaningful patch of ground. A clear liability rule, one that says plainly who pays when an AI gets it wrong, would change how companies sell their products here.

Speech made. Rules still to write.