How Australia uses Claude: Findings from the Anthropic Economic Index

Anthropic is expanding to Australia. We’re opening a new office in Sydney in the coming weeks, and we’ve signed a Memorandum of Understanding with the Australian government to cooperate on AI safety research and support the goals of Australia’s National AI Plan. To mark the occasion, we thought we’d look more closely into how Australians are using Claude.

Australia accounts for 1.6% of global Claude.ai traffic, ranking eleventh among all countries in our February 2026 sample (Figure 1). Its Anthropic AI Usage Index (AUI) is 4.1, which means Australians use Claude over four times more than its working-age population would predict. This places Australia among the highest per capita adopters of Claude with ranking seventh behind Singapore, Israel, Luxembourg, Switzerland, the United States, and Canada.

Within Australia, usage is concentrated in the most populous states. New South Wales accounts for 37.2% of conversations, followed by Victoria at 30.8%, Queensland at 17.7%, and the remaining states and territories at a combined 14% (Figure 2). States and territories’ working age population size explains much of this ordering. However, adjusting for this, New South Wales has an AUI of 1.20 and Victoria an AUI of 1.19. These are the only two states with higher than expected per capita adoption; every other state or territory has an AUI below 1, with adoption lowest in Western Australia at 0.68, Tasmania at 0.32, and the Northern Territory at 0.12.

This pattern for Australia is consistent with the relationship we documented in our economic primitives report between per capita income and use cases: richer countries tend to have lower coursework shares and higher personal use shares.

The broader diversity of usage within Australia comes mostly from a lower share of coding-related tasks. Figure 6 shows how the composition of Australia's top 100 tasks differs from the global top 100 by SOC major group (broad occupation categories defined by the US Bureau of Labor Statistics). Computer and Mathematical tasks are 8.0 percentage points lower in Australia. No other category of use approaches that magnitude of difference to the global average, though education-related tasks are nearly three percentage points below global parity. The offsetting positives are spread across many groups, led by Management (+2.3pp), Office and Administrative Support (+1.3pp) and Life, Physical, and Social Science occupations (+1.3pp).

At the level of individual request clusters—categories of what users ask Claude to do—the pattern is similar (Figure 7). The cluster that is most underrepresented in Australia relative to the global distribution is general coding assistance, at 13.5% of Australian use compared to 16.8% globally. Document translation is also underrepresented, consistent with Australia being a predominantly English-speaking market. The most overrepresented clusters are a mix of personal use (such as personal life management (+1.9pp) and health and well-being support (+1.8pp)), and non-technical professional work, including workplace correspondence (+1.7pp), business documents (+1.6pp), and financial guidance (+1.3pp).

Higher-adoption countries generally use Claude more broadly, so part of Australia's lower coding share is expected. Australia's Computer and Mathematical shortfall of 8.0pp sits right in line with the Anglosphere average of 8.9pp. Where Australia diverges from its peers is in what fills the gap: it leans further into Management (+2.3pp vs the Anglosphere's +1.2pp) and Office and Administrative Support (+1.3pp vs +0.1pp), rather than Educational Instruction (−2.7pp vs +1.6pp for peers).

In contrast to the strong association between income and Claude usage per capita we observe across countries , income does not appear to predict adoption across Australian states and territories (Figure 3). That said, since we can only compare across the eight, the lack of correlation between income and usage is suggestive rather than conclusive.

Local workforce composition is a likely factor for why adoption appears decoupled from income across Australian states and territories: mining-heavy Western Australia, the state with the highest GSP per capita, has low usage per capita. The Northern Territory likewise has high GSP per capita and low AUI. This likely reflects these sparsely populated states and territories having a relatively smaller share of workers in occupations where Claude tends to be adopted. The Australian Capital Territory has above average income but lower-than-expected Claude usage per capita, which may reflect barriers to adoption among its large public sector workforce. The two states with the greatest adoption, New South Wales and Victoria, have slightly below average income; high adoption in these states likely reflects the higher share of workers in finance, professional services, and tech sectors, where Claude usage tends to be higher.

Our fourth Economic Index report introduced four economic primitives: the use-case mix, the degree of autonomy afforded to Claude, task success, and task complexity. Figure 4 shows where Australia falls on each of these relative to other Anglosphere countries (in orange), other high-adoption economies (in grey), and all countries in our sample (in purple).