Anthropic CEO Dario Amodei has been in Australia this week, warning policymakers about the risks of AI and sharing a blunt calculation that the benefits won’t be fairly distributed, while also working to sign agreements on Australia’s capacity to support Anthropic’s needs.
The co-founder behind the powerful AI, Claude, held speed-networking-style briefings with policymakers and answered questions in front of a roomful of around 150 people at Parliament House, for the Anthropic Futures Forum on Wednesday.
His meetings with Prime Minister Anthony Albanese and Treasurer Jim Chalmers saw Anthropic sign a memorandum of understanding under Australia’s National AI Plan. The MoU covers data centres, a commitment to provide workforce impact data and collaborate with the AI Safety Institute, and a promise from Anthropic to provide $3 millon in research credits to Australian universities. There are no enforceable oblihgations of penalties under the MoU, nor is copyright covered.
Given the consequences of AI companies like Anthropic on the future of all Australians, it was Amodei’s broad statements on the inevitability of AI-displacement and the work that will need to be done to address growing inequality as a result that should be raising concerns.
Amodei declared during the moderated forum that “the technology is moving faster than we’d like it” and that “we have to act, but we’re not sure how to act.” Reassuring, coming from the co-founder of one of the biggest players in the AI space.
“I’m probably the one in the industry who talks most about the risks of the technology,” he said. “I tell people: no, it’s not your imagination. These labour displacement effects are serious. The cyber risks are serious. The bio risks are serious.”
Again, all very reassuring. And to be fair to Amodei, he is one of the leading voices on AI risks, and Claude is fast emerging as a more ethically comfortable tool than its main competitor ChatGPT.
Amodei does not believe AI economic growth will be shared broadly, “because a lot of it is capital driven, a lot of it is machine driven, it won’t go to everyone.”
He said the benefits “may be more concentrated than they should be” and that he believes governments need to target taxes on those making the most money from AI.
Tax reform, as we’ve seen in Australia, is not exactly known for being swift in deployment, nor is it easy to get consensus on.
But there’s more.
Amodei also raised concerns about an “all-seeing state” and the risks of AI development in autocracies, where every conversation and camera footage could be used to identify dissent. There are also concerns about military power as a result of AI — although, in better news, there is a potential “renaissance of democratic institutions” coming our way.
From various reports on Amodei’s address (Women’s Agenda was not in the room), the CEO didn’t go into much detail on the gendered impacts of AI rollout, especially regarding jobs displacement. However, just recently Anthropic released research predicting labour market impacts of AI based on early evidence, and found (in Anthropic’s own words) that “workers in the most exposed professions are more likely to be older, female, more educated, and higher-paid.”
As we recently shared regarding how the great AI displacement has arrived and why women’s jobs are most at risk, the UN Women’s Gender Snapshot 2025 found that women are almost twice as likely as men to work in jobs that are at high risk of automation, with 4.7 per cent of working women affected compared to 2.4 per cent of working men. Further research finds that women’s jobs in wealthy countries are three times more likely to be automated than men’s, suggesting the AI labour shock will hit women first.
Still, there are other benefits to speak of, especially in health, with Amodei suggesting that ten years from now, “cancer could look like bubonic plague – something we study more in history books than we deal with today.”
Just what will be the state of our mental health by 2036? Along with our financial and social health?
At least we’ll have data centres. Lots of them. And once those data centres are built, we won’t even need anyone to work in them.
Read our recent piece on the gendered impacts of AI displacement.