Yoshua Bengio warned Australians about the risks of AI technology in a The Australian article published on 23 September 2026, saying threats that were once theoretical are now happening. That warning landed the same day The Australian reported that data centres should not push up power bills for Australian families under national cabinet-endorsed standards aimed at preventing higher household electricity costs.
Who Gets the Warning, Who Gets the Bill
Bengio’s warning puts the danger in plain sight. What was once treated as a theoretical threat is now, in his words, happening. The article did not soften that message, and it did not pretend the people most exposed to the consequences are the ones making the decisions. They’re not. The machinery gets built elsewhere. The risks get spread downward.
The second article, also published on 23 September 2026, shifts from existential fear to the more familiar arithmetic of power: who pays when infrastructure expands. The Australian reported that data centres should not push up power bills for Australian families under national cabinet-endorsed standards. That’s the promise. Keep the costs from landing on households while the AI boom rolls forward.
The wording matters. “Should not” is not the same as won’t. It’s a policy boundary drawn by the same institutional order that keeps expanding the systems ordinary people are expected to absorb. The national cabinet-endorsed standards are presented as a safeguard, but the basic structure remains intact: the growth of AI infrastructure, the pressure on energy systems, and the need to stop families from footing the bill.
The Apparatus Talks Safety
The Australian framed the two stories through separate lenses, but they sit in the same power arrangement. One article focused on existential risk to humanity. The other focused on keeping electricity bills stable as AI infrastructure grows. That split tells its own story. The public gets warned about catastrophe while being reassured that the costs will be managed. The system keeps moving. The language changes.
Bengio’s warning about threats that were once theoretical now happening is the sharpest line in the material. It points to a technology moving faster than the people affected by it can control. The article gives no sign of public control, only expert warning. That’s how these systems usually speak: first through specialists, then through standards, then through the quiet expectation that everyone else will adapt.
The energy story carries the same logic in a different register. Data centres should not push up power bills for Australian families, the paper reported, under standards endorsed by national cabinet. The families are named because they’re the ones who’d pay if the arrangement goes the usual way. The institutions are named because they’re the ones setting the terms. The people at the bottom are told the arrangement is being handled.
What They Call Protection
National cabinet-endorsed standards are the only policy response described in the source material. No grassroots response appears in the articles. No mutual aid network, no community control, no direct action. Just the familiar top-down promise that the apparatus can regulate its own expansion without dumping the costs onto ordinary people.
That’s the whole game in miniature. AI is discussed as a threat to humanity, then as an energy problem to be managed. The public gets the warning and the reassurance at once. The institutions keep their hands on the switches. The families, meanwhile, are expected to trust that the standards will hold.
The Australian published both stories on the same day, and together they show how the AI boom is being narrated from above: existential risk on one side, cost containment on the other. One story warns of danger to humanity. The other tries to keep the electricity bill from spiking. Between them sits the same hierarchy, making decisions, setting standards, and asking everyone else to live with the consequences.