Australia's public dressing-down of OpenAI over a security breach this week has put artificial intelligence regulation firmly back on the global agenda. And the pace of change is striking: it took two decades of mounting harm before governments seriously moved on social media. With AI, the public appetite for controls has arrived in just a few years.
Prime Minister Anthony Albanese has made a habit of taking on big tech. His government was the first to ban social media for young people. It has forced Meta, Google, TikTok and LinkedIn to pay local publishers or face a levy. Now Canberra is pushing a digital duty of care bill that would make platforms responsible for the harms they enable.
The tech giants have pushed back hard. And with the Trump administration in Washington, they have powerful friends. The US Embassy in Canberra this week publicly warned the duty of care bill could be seen as facilitating censorship, a rare intervention in another nation's domestic policy.
Why AI regulation is moving faster than social media rules
On AI, however, Australia is far from alone. This week it was one of 22 middle powers to sign a statement calling for global controls on artificial intelligence. That echoes what the leaders of the frontier AI labs, including OpenAI's Sam Altman, have been saying themselves.
Even Donald Trump, who has mocked globalist schemes to control AI, has conceded some care is needed. His Treasury Secretary, Scott Bessent, has reportedly been exploring a notification system on AI safety with China's Vice-Premier.
The contrast with social media is stark. For years, the debate was stuck on declarations of concern and generic calls for action. With AI, the window between public alarm and political response has compressed dramatically.
What could AI regulation actually look like?
Concrete rules do not exist yet. Early efforts have focused on physical infrastructure, like data centres, rather than the technology itself. The next step, according to leaders like Altman and Bessent, is to start describing the actual shape regulation could take.
The Trump administration has shown that traditional rule-making processes can be bypassed when the will is strong enough. That cuts both ways: it means progress can be fast, but it also means the guardrails could be built by whoever acts first.
Could New Zealand step up on AI safety?
Among the middle powers calling for AI guardrails were Norway, Finland and Singapore, all nations with populations similar to New Zealand's. The opportunity is there for a smaller country with the right ideas to shape the global conversation.
New Zealand has a natural connection to this world. Shane Legg, co-founder of Google DeepMind and a prominent voice in AI safety, went to Rotorua Lakes High School and studied at the University of Waikato. If anyone has the credentials to put a workable plan on the table, it could be him.
The world is looking for someone with the outlines of a workable safety plan to raise their hand. There is no reason that voice could not come from here.
Frequently asked questions about AI regulation
Why is AI regulation moving faster than social media regulation?
Public concern about AI has grown rapidly because the technology is seen as more consequential than social media. Governments are also learning from the social media experience, where delayed action allowed harms to accumulate over two decades.
What are middle powers doing about AI?
Twenty-two countries, including Australia, Norway, Finland and Singapore, have signed a statement calling for global controls on AI. They are pushing for international coordination on safety standards and notification systems.
How could New Zealand contribute to AI safety?
New Zealand could follow the example of other small nations by proposing concrete regulatory frameworks. It also has a direct link to AI leadership through Shane Legg, the Google DeepMind co-founder who was educated in Rotorua and Hamilton.