Key Takeaways
- Wellington Mayor Andrew Little revealed “large chunks” of a $435,000 Deloitte report for the City Council were AI-generated, with errors including double-counted staff and $21.5 million in overstated expenses.
- The “Future Fit Pōneke” report’s own fine print acknowledged its assumptions needed validation and should not be relied upon for decision-making, yet Deloitte has not offered Wellington ratepayers a refund.
- A parallel Deloitte AI controversy in Australia ended with a partial $97,000 refund on a $440,000 contract after fabricated references and a false judicial quotation were found in a government report. Wellington’s City Council paid $435,000 for a Deloitte consulting report that Mayor Andrew Little says contained “large chunks” written by AI, double-counted staff and overstated expenses by $21.5 million. Little made the disclosure on September 2, 2026, and it was not Deloitte’s first time at this particular rodeo.
What the Report Got Wrong
Released in November 2025, “Future Fit Pōneke” initially claimed Wellington City Council employed 330 excess staff. That figure quickly unravelled. Deloitte had counted both occupied and vacant roles in its total headcount, artificially inflating the surplus. The firm also reportedly drew on three-year-old employment data sourced from the Taxpayers’ Union rather than current council figures.
The financial analysis was no more reliable. The report overstated staffing expenses by at least $21.5 million, in part by calculating all employees at a full-time salary regardless of actual hours worked. It also flagged any role containing the word “project” in its title as a duplicate, apparently without examining what those roles actually did.
Public Service Association National Secretary Duane Leo called the report a “flimsy PowerPoint presentation” that “lacks any depth, rigour or even a basic understanding of what the Council’s role is.” Leo also noted that the report’s own fine print acknowledged its assumptions required validation and should not be relied upon for decision-making, a disclaimer that sat awkwardly alongside Deloitte’s headline recommendations.
Leo raised a further concern: Deloitte’s assumption that AI could deliver productivity gains of up to 50% was offered as justification for staff reductions, with no evidence or methodology for how replacing experienced workers with AI tools would achieve that result. Leo described the combination of unproven AI assumptions and recommended cuts as a “recipe for disaster” for vulnerable residents who depend on face-to-face services, and called for the report’s rejection. That critique lines up with St. Louis Fed research published July 31, 2026, which found that roughly 95% of AI-related productivity statements on corporate earnings calls are forward-looking claims about future gains rather than realised results, a gap the PSA argued carried direct consequences when those assumptions were used to justify workforce reductions in a $435,000 report.
The Australian Precedent
In October 2025, Deloitte submitted a 237-page report to the Australian federal government, also valued at $440,000, that was later found to contain fabricated academic references, non-existent footnotes and a false quotation attributed to a Federal Court judge. The AI tool used was reportedly Azure OpenAI GPT-4o, according to reports. Following public and academic scrutiny, Deloitte Australia agreed to a partial refund of approximately $97,000. The partner responsible for the report is reported to have left the firm.
Reviewers of the Australian incident argued that GPT-4o did not malfunction and that the failure lay in Deloitte’s review process. The same assessment applies to Wellington: errors in headcount, salary calculations and role classifications passed through delivery to a government client without correction.
Two Countries, Two Outcomes
In Australia, Deloitte issued a partial refund and a partner departed. In New Zealand, Wellington’s mayor has reportedly been ignored in calls for rectification, and no reduction in fees has been offered. Deloitte has stated it “uses a range of tools and technologies across its work, including AI-enabled tools,” but has not publicly addressed the specific errors in the Wellington report, according to reports.
For public sector procurement teams beyond Wellington, the practical lesson is contractual. Where AI is used in government consulting work, contracts without explicit disclosure requirements, quality-control obligations and defined remedies for material inaccuracies leave clients with limited recourse when errors surface. The Deloitte case illustrates what that gap looks like when a $435,000 report goes wrong and no refund follows.
Originally published at https://autonainews.com/wellington-ratepayers-fund-435000-deloitte-report-mayor-claims-ai-wrote/
Top comments (0)