When public money backs tech, Canberra must insist on local benefits — Harrison.ai is the test case

A year after the federal government invested $32 million to keep Harrison.ai in Australia, the company is laying off staff and pivoting to a US business model. That sequence is a blunt reminder that bold industrial bets in fast-moving sectors carry real risks for taxpayers, workers and local innovation ecosystems. It also raises questions about how governments should design interventions to capture promised national benefits when companies’ strategies change. Governments rightly want to back homegrown tech as part of an economic transition to higher-value industries. The AI sector promises substantial gains: analysts say the economic benefits of AI adoption are beginning to flow in Australia, even as projections indicate significant workplace change in coming decades. But public investment must be more than a headline; it should be structured so taxpayers share upside and downside in a fair and transparent way. The stark fact here is that taxpayer money was deployed with the explicit aim of keeping the company in Australia. The outcome — staff cuts and a shift to the US — will feel like a broken promise to many, especially the employees directly affected and the communities around them. Beyond immediate job losses, there are longer-term implications: a loss of tacit knowledge, a hole in the local talent pipeline, and a signal to other founders and investors that government dollars do not guarantee anchoring of industry. This is not an argument against public support for innovation. It is an argument for smarter support. Conditionality, milestone-based funding, clawbacks where public investment fails to produce committed local outcomes, and clear transparency about the terms of deals should be non-negotiable. Governments must design interventions that protect community interests while remaining nimble enough to back risky ventures. Equally important is a sober national conversation about what success looks like. Is it a handful of headline-grabbing start-ups, a broad-based build-out of mid-sized firms employing Australians, or widespread skills upgrading across industries? Policy choices should be guided by those objectives: tailored support for training, stronger commercialisation pathways for research, and incentives that favour local re-investment. There is also a reputational cost. When public investment does not yield durable local benefits, it undermines public confidence in industrial strategy and feeds scepticism about the government’s competence. That sentiment can make future interventions politically harder to mount, at a time when many economies face genuine structural change driven by automation and AI. For workers, the human cost is immediate. For taxpayers, the question is whether the public purchase delivered value. For policymakers, Harrison.ai’s pivot should be a prompt to refine the instruments used to back innovation: more rigorous contracting, faster evaluation, and a clear line between public risk-taking and private reward. If Australia is to secure the dividends of AI, it must learn to do public backing better. That means not stopping at the cheque book, but building frameworks that lock in lasting local benefits — for jobs, skills and the wider community.