What is Anthropic’s first Australian federal contract?

Anthropic’s first direct contract with the Australian federal government is with the Fair Work Commission, granting access to top-tier Claude Opus models. It was signed on 19 March 2026.

Why did the Fair Work Commission choose Anthropic?

The Commission’s workload is on track to be over 70 per cent higher this financial year than three years ago, with President Justice Hatcher attributing the growth principally to litigants using generative AI. It released draft guidance regulating AI use in tribunal cases on 24 March 2026 and awarded Anthropic the contract eight days later, making it the first Australian tribunal to formally regulate AI submissions and the first federal agency to award Anthropic a direct contract.

Does the Anthropic MoU give Anthropic any procurement advantage?

No. A spokesperson for Science Minister Tim Ayres told The Canberra Times any government-wide integration of Anthropic products “will be in line with the APS AI Plan and follow standard Commonwealth procurement processes.” The MoU is non-binding and covers AI safety collaboration, Economic Index data sharing, A$3 million in research API credits and infrastructure alignment. Each agency procures AI tools under its own competitive process.

How does Anthropic’s Australian federal entry compare to OpenAI’s?

OpenAI signed its first Australian federal contract in June 2025, a A$50,000 limited-tender deal with Treasury, and in October 2025 added a A$25,000 contract with the Commonwealth Grants Commission. Both sat below the A$80,000 competitive tender threshold. Anthropic’s first direct contract is with the Fair Work Commission, an operationally heavy quasi-judicial tribunal. Both companies pair federal procurement with active Australian data centre infrastructure programmes.

What is the Sydney hiring pattern around the federal contract?

Anthropic’s Sydney Compute team (its in-house data centre function) has grown from four open roles on 18 May 2026 to six on 28 May, now 35 per cent of the 17 open roles in that function worldwide. A new Sydney Account Executive, Public Sector role is open to lead the company’s market entry across Australian government departments. Sydney also carries the Head of ANZ Applied AI, the ANZ Communications Lead, the Head of Enterprise Sales Industries ANZ, and country GM Theo Hourmouzis, named in early May 2026.

Anthropic signs first Australian federal contract with Fair Work Commission

At a glance

Anthropic’s first direct contract with the Australian federal government is with the Fair Work Commission, granting access to top-tier Claude Opus models.
The contract was signed on 19 March 2026, Anthropic and the Australian Government signed the National AI Plan MoU a few weeks later on 31 March 2026. The MoU is non-binding and a ministerial spokesperson stated that any APS-wide integration “will follow standard Commonwealth procurement processes.”
The Fair Work Commission’s selection is consistent with a 70 per cent workload increase in three years, which President Justice Hatcher has attributed principally to litigants using generative AI to prepare applications and appeals.
Anthropic’s Sydney Compute team has grown from four open roles on 18 May 2026 to six on 28 May, now 35 per cent of Anthropic’s 17 global Compute postings. A new Account Executive, Public Sector role is open in Sydney with the remit to lead market entry across Australian government departments.

What the contract covers

The Fair Work Commission contracted Anthropic for software support and access to Claude-based AI services under a direct agreement executed on 19 March 2026 in Canberra. The $13,358 agreement runs for 12 months and falls under software maintenance and support, indicating access to Anthropic’s AI models alongside ongoing technical support rather than a one-off deployment.

The contract was awarded through a limited tender with a single supplier invited, suggesting the Commission had already identified Anthropic as the preferred provider. While no formal justification was published, a similar contract awarded to the Australian Signals Directorate cited a lack of equivalent capability from Australian or New Zealand vendors, pointing to a broader reliance on US-based frontier AI providers.

A few weeks later on 31 March, Anthropic CEO Dario Amodei met Prime Minister Anthony Albanese and Science Minister Tim Ayres in Canberra to sign the National AI Plan MoU. A spokesperson for Minister Ayres told The Canberra Times that any government-wide integration of Anthropic products “will be in line with the APS AI Plan and follow standard Commonwealth procurement processes.” The Fair Work Commission contract is a separate procurement decision by the agency itself, and appears to reflect early-stage, agency-led experimentation with generative AI tools rather than a coordinated government rollout.

How the FWC came to be the first agency

The Fair Work Commission has been simultaneously regulating AI use by parties before it and adopting AI for its own operations. On 18 February 2026, FWC President Justice Hatcher told the Victorian Bar Association that the tribunal’s workload was on track to be “over 70 percent” higher in this financial year than three years ago, attributing the growth “principally” to litigants using AI tools. Hatcher described running his own experiment with ChatGPT, which produced an unfair dismissal application and witness statement in under ten minutes, inventing facts and suggesting compensation quantum.

On 24 March 2026, five days after the Anthropic contract was signed, the Commission released its draft Guidance Note on the use of generative AI in Commission cases for public comment, with submissions due by 4pm AEST on 10 April 2026. The draft sets three requirements: disclosure of GenAI use in any submitted document, verification of all facts and authorities, and a recommendation that GenAI not be used to author the substantive content of witness statements or declarations.

The Commission is the first Australian tribunal to formally regulate AI submissions in its proceedings, and the first Australian federal agency to award Anthropic a direct contract. The two facts read together describe an agency that has decided AI is now load-bearing for both case management and case adjudication.

What the contract enables in practice

The contract gives the Commission access to Claude Opus, Anthropic’s top-tier model, with no public detail on which use cases it has been scoped to. The workload pressures Hatcher described point to four likely applications: triage of incoming applications so meritorious cases reach a member faster and AI-padded ones get flagged sooner; summarisation of long submission and evidence bundles; drafting of routine procedural decisions, directions and form correspondence; and case-law and award lookups against the Commission’s own corpus. A working deployment would help the tribunal close the gap on its 12-week decision target (currently slipping) and give every other agency under the APS AI Plan a procurement template to reference.

The Sydney team being staffed around the federal pipeline

When we covered Anthropic’s Sydney data centre operations hiring on 18 May 2026, the company had three new Sydney Compute roles open (Facility Operations Lead, Hardware Operations Lead and Senior Data Center Capacity Delivery Manager AUS) plus a Transaction Principal live since March. Compute is Anthropic’s in-house data centre function, modelled on the US team that runs capacity built under the company’s US$50 billion Fluidstack partnership.

The Sydney Compute roster has grown in the ten days since.

Sydney Compute roster	18 May 2026	28 May 2026
Transaction Principal	Open	Open
Senior Data Center Capacity Delivery Manager, AUS	New	Open
Data Center Facility Operations Lead	New	Open
Data Center Hardware Operations Lead	New	Open
Data Center Electrical Engineer	Not posted	New
Data Center Mechanical Engineer	Not posted	New
Sydney share of global Compute postings	4 of 17 (24%)	6 of 17 (35%)

Source: Certified Strategic Editorial, primary review of anthropic.com/careers/jobs, 18 May 2026 and 28 May 2026.

The Sales function has moved in parallel. A new Sydney Account Executive, Public Sector role requires 10 or more years of enterprise sales experience in the Australian public sector, deep understanding of Australian government procurement processes, and Australian security clearances. The role “will lead Anthropic’s market entry and expansion in the Australian public sector, establishing our presence and building relationships with key stakeholders across government departments and agencies.” That is the procurement-facing role that converts contracts of the Fair Work shape, and it is being hired now.

Sydney also carries the Head of ANZ Applied AI, the ANZ Communications Lead, the Head of Enterprise Sales Industries ANZ, and an Enterprise Account Executive (Industries) role, under country GM Theo Hourmouzis, named in early May 2026. The hiring depth is consistent with a country at disclosed-capex stage.

OpenAI’s parallel federal expansion

OpenAI signed its first Australian federal contract in June 2025, a A$50,000 limited-tender engagement with Treasury for “software as a service”, followed in October 2025 by a A$25,000 contract with the Commonwealth Grants Commission for “provision of AI”. Both deals sat below the A$80,000 procurement threshold that triggers competitive tender requirements, and OpenAI was the only company invited to bid. The InnovationAus piece on the Fair Work Commission contract notes OpenAI is deepening one of its existing Canberra contracts on the same week the Anthropic deal became public.

What to watch

The FWC Anthropic Contract Notice on AusTender. Any contract above A$10,000 (including GST) must be published; the value, term and procurement method will surface once lodged.

The next APS agency. The APS AI Plan, released in November 2025, requires departments to appoint a Chief AI Officer and to streamline AI procurement. Anthropic’s Sydney Public Sector AE hire is a leading indicator that more agency contracts are in the pipeline.

The GovAi positioning. The government’s in-house ChatGPT-style engine, GovAi, had only 7 per cent regular usage across the APS in its first year. Agencies licensing third-party frontier models in parallel will need to articulate how those deployments sit alongside GovAi, particularly where data sovereignty, classification handling and procurement justification are in question.