16/10/2025
Understand the cost before paying the price.
Deloitte’s recent refund to the Australian government over a $440,000 report riddled with AI-generated errors is more than a cautionary tale—it’s a wake-up call for AI users in business.
The report, intended to assess welfare compliance systems, included, hallucinated evidentiary references, and undisclosed use of generative AI.
While Deloitte claims the “substance” remained intact, the damage to trust and credibility is clear.
The risks of AI misuse is amplified in New Zealand. Government reports and policy frameworks increasingly intersect with Māori data, tikanga, and Te Tiriti o Waitangi obligations, When AI tools hallucinate facts or misrepresent legal precedent, they don’t just produce sloppy work—they risk silencing Indigenousa voices and distorting culturally grounded knowledge.
Imagine an AI-generated report on Māori health, education, or justice—built on flawed data, missing nuance, and lacking consultation. The consequences are systemic and deeply harmful.
This case raises urgent questions for our public sector:
• Are AI tools being used in ways that uphold Te Tiriti principles?
• Is there transparency when AI contributes to policy or research?
• Who is accountable when AI-generated content misrepresents Indigenous knowledge?
As Aotearoa embraces digital transformation, MahiAi. ensure AI is used ethically, transparently, and in partnership with tangata whenua. The Deloitte case shows what happens when shortcuts replace expertise—and why Indigenous voices are central to successful AI governance.
Let’s not wait for a refund to realise the cost of getting it wrong.
The Guardian | Krishani Dhanji