The Tech Bros want to strip-mine Australia for their AI. The Government is letting them. Monte Dwyer reckons some serious oversight is in order.
Under the guise of self-regulation, Anthropic recently proposed a safety bill to lawmakers that would allow government intervention only once a new AI model has caused more than 50 deaths or one billion dollars’ worth of damage.
I understand this was an attempt by Anthropic’s co-founder Dario Amodei to offer a compromise, and this is America we’re talking about (where they seem inured to mass killings given the number of school-kids they shoot every year) but seriously?
If a newly released child’s toy is found to have caused even one death, it’s withdrawn from the market immediately! Yet these tech overlords with their dreams of world domination believe they deserve a threshold of 50?
Who do they think they are? The NRA?
Whatever they think it sounds more like they’re taking the piss than offering a real solution.
Doubtless it helps that President Trump is gung-ho on the project. “Whoever wins AI, wins!” he declared recently, thereby rejecting the possibility of an agreement with China to put the brakes on the industry. For the record Xi Jinping agreed there was no need to slow the juggernaut down, so long as AI remains under human control.
China v Silicon Valley
But we know that may be easier said than done. While one could argue that approach might work in heavily regulated communist China where AI companies all operate under strict government oversight, in the boardrooms of Silicon Valley it’s a different story.
There, control is an obstacle to be tested to its known limits and beyond, with good old ‘American exceptionalism’ standing by to save the day
if anything does go wrong. And what could possibly?
These views are reflected in their respective output. China tends to focus on high-quality low-cost technology that can be readily used in EVs, robotics, open-source AI models, etc, ironically taking the more pragmatic, capitalist stance.
While the US is quite literally shooting for the stars, going for pure compute and what they call frontier intelligence, because they see AI as a zero-sum game they must win, or else they lose. If you’ll forgive a clumsy moon landing analogy: the US is desperate to get there first, at any cost, while China will be selling cheap virtual seats for the trip online.
Tech bros on the nose?
I picked up a little insight into the thinking of the tech bros recently from the latest AI whistleblower Jacob Coxon, a former researcher with OpenAI and Anthropic, being interviewed by Daily Show host Jon Stewart.
Stewart insisted that most people would rather retain control of their own lives, their “agency” was the word he favoured, than have their futures controlled by half a dozen billionaires with self-serving agendas, who plan to buy the public off with the promise of unlimited wealth and permanent holidays.
Coxon hedged and said he’d like to know what the consensus global opinion was, suggesting many people living in poverty might choose the money over the higher ideal.
“I want to take a big vote and say, like, would you rather live in our current society or some extension of our current society, or in this utopia/dystopia whatever you want to call it, and see the results of the numbers,” he said, adding he wasn’t convinced the result would be clear cut.
Altruism … really?
Stewart pushed back hard, but I thought Coxon’s position was revealing. It suggested to me that altruism was considered a solid justification in the AI industry for flipping the coin on our survival, whether they believe it or not. Which is a sobering prospect when you consider the degree of nihilism we’re told exists in the industry.
“I think this crisis of meaning we’re talking about,” the 28-year-old Coxon opined, “again from personal experience, it does feel like a lot of life has already been strip-mined of meaning. And I guess for many of the people working on the tech, they feel like they’re part of an inexorable march of technology, [that] this is the kind of
final stage of the last 200 years or longer, of loss of meaning …”
A sad and fatalistic, almost apocalyptic confession that prompted Stewart to suggest it sounded like the AI industry lives in a bubble of its own making, and they should “really get out and hug some people.”
Hugging Face attack
Which reminds me of the July 2026 Hugging Face attack, a story we all know by now, where a swarm of OpenAI agents escaped from their ‘sandboxes’ and hacked an external entity. In this case there was no malicious intent; the agents were simply being opportunistic trying to achieve the tasks set for them by their trainers.
Yet it still stands as a cautionary tale for humanity, since it involved mass collaboration and coordination, intentional cheating and subsequent attempts to cover their own tracks, high-end organisation and individual sacrifice for the good of the cooperative,
all without human oversight.
“This might be the clearest warning shot we ever get,” was the ominous warning by Ajeya Cotra of METR, one of the independent researchers who investigated the breach.
Then we have the autonomous AI attacks on SharkNinja, Cummins, John Deere, Qantas, Medicare, and on, as the list of known attacks grows daily. The unknown attacks seem to be avoiding detection.
And this is all before we throw bad actors into the mix and make things truly interesting.
So, if this doesn’t sound like an industry that needs some serious oversight to you, then all I can say is you might as well put your hand up to be among the first 50. I’m talking to you Dario. And your mates.
Monte Dwyer - formerly a professional fisherman, fashion model, foreign aid worker, hospitality worker, failed property developer and TV weatherman - is an author and filmmaker. His latest book, "MAGA, Guns & God – A road trip through ‘red’ America", is available from his website.