Five days after a federal judge blocked the Pentagon from adding Anthropic to a supply-chain risk blacklist, CEO Dario Amodei was on a chartered jet to Canberra.
He landed with three Australian lobbying firms already hired, a commitment to cover full grid upgrade costs in any Australian market where Anthropic builds, and a full political schedule: Treasurer Jim Chalmers, Industry Minister Tim Ayres, and Assistant Technology Minister Andrew Charlton. He met Prime Minister Anthony Albanese the following day to sign a memorandum of understanding on AI safety and research, according to Anthropic's announcement. The MOU covers AI evaluation, red-teaming, and responsible development — phrases that land differently when you know what Anthropic refused to let the U.S. government do with Claude.
Anthropic told the Pentagon it would not allow Claude to be used for autonomous weapons or mass surveillance of American citizens, according to NPR. The Trump administration added the company to a supply-chain risk list anyway. Anthropic filed two federal lawsuits in March alleging illegal retaliation. A federal judge granted a preliminary injunction blocking that designation on March 26, CNBC reported. Five days later, Amodei was in Australia.
The MOU is the output. The lawsuit is the explanation for why the trip was worth making.
The Albanese government has no binding AI legislation. It rejected a text-and-data-mining copyright exception in October 2025 that would have allowed AI companies to train freely on Australian works without licensing them individually, the National Law Review reported. Australia is not rolling out a regulatory red carpet. It is making a deliberate policy bet on Anthropic specifically, at a moment when the company's U.S. relationship with the Defense Department is in active litigation.
Anthropic told an Australian Senate committee it would cover the full cost of grid upgrades and bring net-new power generation online wherever it builds, the Sydney Morning Herald reported. AI data centers are grid-intensive. Australia has renewable energy ambitions and an electricity grid under pressure from manufacturing growth and electrification. The offer to pay for upgrades and new generation addresses the infrastructure question every government asks before welcoming a hyperscaler.
Anthropic is also opening a Sydney office, its fourth in Asia-Pacific, according to Anthropic's MOU announcement. It has hired three lobbying firms: Anacta Strategies, SEC Newgate, and Carolyn Hough Policy Australia, the Sydney Morning Herald reported. During the Canberra visit, Anthropic demoed its AI agents to public servants and government officials at a Futures Forum inside Parliament House, a pitch to the people who will shape Australia's AI policy, not just the politicians who signed the MOU.
The company, valued at US$380 billion (about A$550 billion), has government partnerships. Since 2024 it has worked with Palantir on national security operations, including rapid data processing, trend identification, and document review, according to NPR. That relationship is still active even as Anthropic sues the administration over a different part of its government work. The company is simultaneously a national security partner and a litigation adversary.
The MOU itself is light on specifics: a shared commitment to AI safety evaluation, some language about responsible development, an intent to coordinate on international standards. It is a letter of intent dressed as a bilateral agreement. What it signals is harder to quantify: a Western democracy with no AI law is making room for a U.S. AI lab that told the Pentagon no, and that lab showed up with a chartered jet, a checkbook for grid upgrades, and a plan to hire locally.
