Anthony Albanese says it’s a lie he doesn’t personally use AI, although for security reasons he cannot say for what. Speaking to The Saturday Paper, the prime minister dismisses Opposition Leader Angus Taylor’s claim that he is out of touch with the technology.“I do use AI. He’s an idiot,” the prime minister said on Thursday, adding: “One of the challenges here is that you need access to artificial intelligence to engage with some of the risks that are there from artificial intelligence.“There are elements of the emergence of AI that are unique, that present challenges that aren’t the same as business as usual. They aren’t the same as the cybersecurity issues that have been an issue for some time.”Albanese says AI companies are very powerful and the technology is moving very, very fast. “Which is why we need a response to ensure sovereignty is protected, and to ensure engagement, and that is what we have been doing with companies like Anthropic and OpenAI.”The prime minister is keen to make clear that he has met with AI executives. Most prominently, he signed a non-binding memorandum of understanding with Anthropic chief executive Dario Amodei on April 1.“I’ve engaged with international partners,” he says. “I’ve met with companies when they’ve been here in Canberra and in Sydney, including Dario and others.”Observers say these meetings point to the level of lobbying being undertaken by AI companies, borrowing from principles established by big tobacco. In Australia, that has involved hiring multiple Labor-linked staff. In America, it has involved multimillion-dollar donations to an already compliant administration.Ed Husic, who as former industry minister led the government’s initial response to AI, says there is a huge asymmetry between the resources of the big AI companies and the expertise of government.“The imbalance is massive and, as a result, it’s like pushing against an open door,” he tells The Saturday Paper. “What we’re witnessing is nothing more than a shock and awe lobbying approach by US tech, designed to swamp government decision-making.”An AI strategist who has worked inside and outside government made a similar point: “The amount of power these people have is incomprehensible … This isn’t just a product that’s being sold. This is dual-use capability. This is the equivalent of a private company developing nuclear weapons.”OpenAI established an office in Sydney last December, although its engagement with the government had started earlier. In June 2025, the company’s chief economist, Ronnie Chatterji, a former Biden economic adviser, led a team that went to Canberra to meet government figures and senior bureaucrats. Chief strategy officer Jason Kwon came to parliament in mid September.Last year, OpenAI hired Kate Pounder to help establish the company in Australia. She left in December. Pounder was the inaugural chief executive of the peak industry body, the Tech Council of Australia, and a former business partner of assistant technology minister Andrew Charlton, who is central to the government’s response to AI.On Thursday, she was appointed as a commissioner to South Australia’s Royal Commission into Artificial Intelligence, along with Reserve Bank board member Iain Ross and Stanford University machine learning expert Christopher Manning.Charlton and Pounder founded AlphaBeta Advisors together, an economics firm sold to Accenture before Charlton entered parliament in 2022. Mandala Partners emerged from the advisory and now publishes AI-focused research commissioned by the tech sector.The Saturday Paper asked Charlton this week about the significant shift in personnel from government to work for the AI companies.“We are focused on Australia’s national interest. It’s not about the personnel,” the assistant minister said.“It’s about making sure that we are clear about what the criteria is for investment. That criteria prioritises Australia and Australians. And if the investment is within that criteria, then we welcome it.”In March, Anthropic engaged Labor-aligned government relations and lobbying firm Anacta Strategies. Anacta’s co-founders, Evan Moorhead and David Nelson, are both former senior Labor strategists. Moorhead was also a former Queensland state Labor MP and state secretary. Anacta directors include former ACT Labor secretary Ash van Dijk and former WA Labor chief of staff Craig Comrie. Anacta also represents social media giant TikTok in Australia.As well as Anacta, Anthropic engaged government relations firm SEC Newgate  at the same time. It also hired AI adviser Carolyn Hough, who was appointed to the attorney-general’s reference group on copyright and AI by the Albanese government in 2024, after serving on the Rudd government’s Ministerial Advisory Consultative Group on Cybersafety.The same month, Amodei met with Albanese and Treasurer Jim Chalmers, just ahead of opening the company’s Sydney office in April.In June, Peter Anstee took up the lead position at OpenAI’s Asia-Pacific national security policy team. Anstee, a former senior adviser to Malcolm Turnbull and a “top talent”, according to the former prime minister, was poached from the Department of the Prime Minister and Cabinet, where he was a first assistant secretary in the department’s national security division. He had previously held a similar position in Home Affairs. Anstee was one of the senior bureaucrats spotted at a Canberra lakeside dinner with Chatterji in June 2025.About the same time as Anstee went to OpenAI, David Masters moved from Australian software giant Atlassian, where he was global head of public policy, to run Anthropic’s government engagement.In July, OpenAI engaged Hanbury Strategy and Communications to lobby on its behalf in Australia. It uses Hanbury’s head office in London. The local Hanbury branch is headed by Owen Torpy, the former chief of staff to Labor minister Amanda Rishworth, and one of its directors is Annabel Clunies-Ross, a former senior adviser to Liberals Josh Frydenberg, Sussan Ley and Angus Taylor.In August, Anthropic appointed its general counsel, former US ambassador to Australia Jeffrey Bleich, as its international special envoy, in what is seen as an Australia-centric move.“He gets Australia,” the AI strategist notes of Bleich. “There’s a lot of diplomacy that’s needed here. He’s a really good person for what is needed and people respect him. The role didn’t exist. It is fascinating.”The key decision-makers in the Albanese government are the prime minister, along with Chalmers and Charlton, who is the junior minister to Industry Minister Tim Ayres.Within the Prime Minister’s Office, the lead advisers are the head of policy, Sam Trobe; former Ed Husic senior adviser and AI ethics expert Ellen Broad; and national security adviser Gemma Smyth.There is also the new Office of AI in the Department of the Prime Minister and Cabinet, headed by Fergus Hanson, which has a coordinating role and is still building staff through secondments from other areas.The Australian AI Safety Institute is headed by philosopher and air force reservist Kate Conroy, who has a concurrent job introducing “responsible AI” in air force systems.“To me it just highlights that vested interests have been having their say behind the scenes.”Until recently, there has been a heavy Treasury lens on AI. The intergenerational report describes AI as having the “potential to drive the biggest economic transformation of our lifetime” and a defining influence that will reshape Australia’s economy over the next 40 years. Treasury, in its submission to the Joint Select Committee on Artificial Intelligence, cites industry estimates suggesting Australia’s data centre pipeline could support more than $150 billion in investment by 2030, the equivalent of 5 per cent of today’s nominal GDP.After the Medicare breach, however, there has been a change.OpenAI has apologised for the breach and said it would “do better for Australia” after it admitted it sent out its agents “without the full set of safeguards used in our publicly available products”.It has committed an expert response team to find and fix weaknesses in Australian agencies and has promised to contribute to efforts to strengthen government cyber defences. Under a directive from Home Affairs, all departments and agencies have been told to take stock of their legacy systems and cyber defences.Albanese regards the company’s response, including directly from chief executive Sam Altman, as a “pretty comprehensive apology” and points to it leading to subsequent admissions of thousands of sites potentially being affected, including United States government sites.Independent Senator David Pocock wants far more transparency in the interactions between government and lobbyists and says the breach shows that better, independent safeguards are needed – and that the government is being pressured covertly. “To me it just highlights that vested interests have been having their say behind the scenes, and the government goes from announcing an AI Safety Act to saying that we don’t need this at all, and our existing regs are fine.”Zeerak Talat, an expert in AI from the University of Edinburgh, says there is a manipulated narrative behind the public statements of AI executives, including after the Medicare hack.“A lot of this is predicated on saying, ‘Hey, we’re the good guys who are doing bad things, so that other bad people won’t do it,’ ” he says. “Like, ‘Look at how powerful these things are. No one can control them. You need us on your side, and you need us to be part of this and to bring us into the government and be another layer in the government.’“The second part is an exceptionalism argument. That AI is such an exceptional tool that we need exceptional laws for it, rather than considering it as: AI is another tool among many that we regulate through existing regulatory measures.“There’s an assumption here that cybersecurity laws don’t apply to them. This also goes for their use of copyrighted material and so on.”The opposition is criticising Labor for being soft on cybersecurity but also says a fair deal for creatives and journalists is possible as the government negotiates a copyright deal to incentivise AI investment in Australia. Angus Taylor says frontier AI is a must for Australia’s cyber defences.According to colleagues, Taylor is immersed in the technology and its implications, bringing expertise as a Turnbull-era cybersecurity minister and a former management consultant.“He’s gone deep on understanding AI and how it works, and he uses it every day,” the Coalition source says.In Taylor’s AI circle are shadow science, technology and innovation minister Aaron Violi, and frontbenchers Jonno Duniam, Andrew Hastie, James Paterson and Michaelia Cash.Within the opposition leader’s office, there’s no specific AI lead, but there is top engagement by Taylor’s head of policy, Steve Hamilton, and chief of staff, Sam Riordan.“People underestimate Angus’s grasp of policy detail, particularly in an area where he’s got experience and a background in [it],” the Coalition source says.“And then his McKinsey days of getting to first principles and really understanding what’s the opportunity, what are the risks, what are we actually negotiating here – he’s very sharp in that space.”Coalition figures say the AI companies know a deal needs to be done in parliament with either the opposition or the Greens for the next big AI steps for Australia, including copyright and national standards.“It doesn’t feel as aggressive or as in your face as people would assume,” a senior opposition source tells The Saturday Paper.“It’s clear that their focus is on the government because they know the decision needs to be made now. You can tell they want to get the government somewhere and then bring us along, but their level is a lot higher.“They’re talking obviously to Angus and they’re talking to the government.”Lobbying and other AI engagement is expected to ramp up in the next few weeks, with the hearings for the AI committees and the copyright consultations continuing. The government is hoping to release still-to-be-determined national standards before Christmas.In Australia, neither Anthropic nor OpenAI have reported donations to the major political parties, although that may change when the Australian Electoral Commission’s transparency portal moves to monthly “real-time” disclosures next year.“I would anticipate at some point they would do what a lot of big players do and probably join the business forums of each party and sort of broadly support both,” one MP quipped.