September 17, 2026 — 3:56pm
OpenAI and Anthropic, the dominant frontier labs, have thrown plenty of elbows in their race toward stratospheric, trillion-dollar stockmarket listings. But this week, Sam Altman and Dario Amodei, their respective leaders, reached a rare unity ticket along with fellow AI kingpin Elon Musk, calling for a slowdown on artificial intelligence development in the name of saving humanity.
This latest bout of fear coming from inside the house began after Anthropic researcher Jacob Poxon resigned last week, accusing the companies of “gambling with our lives”. That announcement prompted his former colleague, Evan Hurbinger, to claim that there was a greater than 10 per cent chance that AI would “kill all humans” within the next decade.
How very reassuring.
Fortunately, all this has been rejected by US President Donald Trump, who has insisted that attempts to slow down AI are a “sick conspiracy” to let China get ahead, creating the absurd situation where highly valuable companies are begging a recalcitrant government to regulate them.
Most people’s concerns about AI are more prosaic. They are less worried about geopolitical rivalry or the prospect of rogue agents creating a devastating virus than whether the machines will take their jobs.
In May 2025, the Anthropic boss Amodei predicted that half of all entry-level white-collar jobs would be obliterated in the next five years, sending unemployment rates into the double digits.
If there is perhaps one micro-dose of optimism among all the bleak AI news, it is that the tsunami hasn’t yet hit. Instead, employment data from around the world paints a messy, paradox-filled picture.
The United States added 162,000 jobs in September, exceeding expectations and strengthening the case for the country’s Federal Reserve to raise interest rates, which it did for the first time since 2023 on Wednesday, much to Trump’s chagrin.
Here in Australia, unemployment crept up unexpectedly to 4.5 per cent in July, its highest rate in the post-COVID era in trend terms. While it’s too soon to tell whether this is the start of a longer-term trend, a 0.1 per cent increase does not seem like the start of a job apocalypse.
The rose-tinted view of the AI boom suggests that, like every past industrial revolution, the latest technological rupture will destroy many jobs but create different new ones. Some recent number-crunching in The Economist magazine found that AI had created around 1 million new jobs in the United States, with many attributable to the data centre construction boom.
The problem is that while data centres are delivering reliable, well-paid blue-collar work, those jobs are temporary. These AI “factories” don’t resemble the real thing when it comes to employment.
Then there’s the fact that they have become a widely despised symbol of everything people resent about AI, with opposition to their development cutting across political lines and threatening to stall that job explosion.
Whether that infrastructure delivers in Australia is also contested. The local sector’s lobbyists insist it will “support” 935,000 jobs. But as this masthead recently reported, that figure comes from a line in a 2024 report by consulting firm Mandala (which has partnered with OpenAI in the past) and conflates data centre work with all tech sector jobs. More importantly, the report was commissioned by five of the country’s biggest data centre operators.
Separately, economic think tank the e61 Institute claims there are only around 11,500 data centres workers in Australia nationally. Even if data centres are the future, it’s quite a stretch to imagine soft-handed software engineers (let alone journalists) to retrain as construction workers.
Beyond blue-collar work, AI is still causing ruptures in the labour market. Research from Princeton University’s Sania Edlich and Apollo Global Management chief economist Tortsen Slok found that rather than destroying jobs, AI was stifling wage growth. Those “different” jobs created by the AI boom could well be repetitive, boring and lower paid. Witness the poor workers recording themselves doing tasks as dull as screwing lids onto milk bottles to train future generations of robots how to hold things.
The next few years could see Australia struggle with the “K-shaped” economy which has characterised the United States post-COVID in which the asset-rich few thrive thanks to a surging stock market and corporate profits, even as wages stagnate for the many.
The problem is that AI is moving at such a fast clip that there are many people with opinions, but few are truly informed. Now, it is the frontier lab bosses, the people who have seen what the technology can do, who are sounding the loudest alarms.
At least the unemployment rate probably doesn’t matter too much if humanity is wiped out.
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
Kishor Napier-Raman is a senior business writer for The Sydney Morning Herald and The Age. Previously he worked as a CBD columnist and reporter in the federal parliamentary press gallery.Connect via X or email.