Only 30% of Australians Think AI's Benefits Outweigh Its Risks

Only 30% of Australians believed the benefits of AI outweighed the risks last year, and in 2026 that trust appears to be falling even further. That is the central tension Edward Santow and Sarah Sacher lay out in a new Guardian comment piece: Australia's two major parties are converging on an AI-friendly policy just as the public runs the other way.

The numbers are stark, and the politics are stranger. Labor and the Coalition, parties that rarely agree on anything, are edging toward what the authors call a "major bet on artificial intelligence." The authors — Santow directs the UTS Human Technology Institute, though the piece itself frames him simply as a co-author — describe this convergence as a "reckless gamble" in their headline. That's strong language for a policy disagreement, and it signals how high the stakes feel to AI governance advocates.

What does the polling actually say?

The piece cites KPMG's Trust in AI global insights survey for 2025. The key figure: only 30% of Australians believed AI's benefits outweighed its risks.

The authors add that "in 2026, community trust in AI appears to be falling even further." No fresh percentage is given for this year, but the direction of travel is the point. Australians are not warming to AI as it spreads into workplaces and public services — they're getting warier.

That matters because AI policy built without public consent tends to collapse at the first visible harm. Takeaway: the public has moved before the politicians have.

Why does the cross-party consensus matter?

According to Santow and Sacher, Labor and the Coalition "increasingly want to take a major bet on artificial intelligence." In Australia's system, when both major parties align, there is effectively no parliamentary opposition to a policy direction. Debate moves to the margins.

The authors argue this consensus is forming at exactly the wrong moment. "Yet, just as the parties are coming together, the Australian community is running the other way," they write. The gap between elite consensus and public skepticism is the story's engine — and it echoes similar gaps seen in other countries debating AI regulation.

A government that bets heavily on AI adoption while trust sits at 30% is betting against its own electorate. Takeaway: bipartisan agreement doesn't equal democratic consent.

What do workers actually want?

The piece opens with a direct claim about public expectations: "Australian workers expect the government to act to protect them from AI harms."

That framing shifts the debate from abstract technology policy to workplace protections. Workers aren't asking whether AI should exist — they're asking who pays when it goes wrong: when a hiring algorithm discriminates, when automated decisions damage livelihoods, when surveillance creeps into the workday.

The authors also insist the path forward isn't theoretical. "There is a clear reform solution at hand," they write — though the excerpt available doesn't specify what that solution is. The full piece, published in the Guardian's comment section on October 21, 2026, presumably details it. What the excerpt does establish is the authors' core position: the problem isn't a lack of options, it's a lack of political will.

Takeaway: the argument is about political choice, not technical possibility.

What's at stake for Australia's AI policy?

Australia has spent years consulting on AI governance without landing comprehensive legislation, even as the EU moved ahead with the AI Act and other jurisdictions tightened rules. The Santow-Sacher piece lands in that context: a moment when both major parties appear ready to lean into AI adoption, and critics fear the safeguards will lag.

The authors' warning is blunt. Headlining the piece "a reckless gamble" sets the terms of the debate they want to have: that accelerating AI uptake on a 30% trust foundation isn't pragmatism, it's negligence.

For readers tracking AI policy globally, Australia is now a test case in one specific question: can democratic governments push AI adoption faster than their publics are willing to accept? If Santow and Sacher are right that trust is still falling in 2026, the collision between the parties' bet and the community's skepticism is coming — and the reform the authors say is "at hand" will either arrive before that collision or after it.

Takeaway: the timing of reform, not its existence, is what Australia's politicians will be judged on.