Errors in Government Report Trigger Ethical Concerns Around AI Use in Consultancy Services
Global consulting firm Deloitte has agreed to partially refund the Australian government after it was revealed that a report the company produced contained multiple errors linked to the use of generative AI tools. The report, part of a government initiative titled “Future Made in Australia”, was found to include fabricated citations and a false legal quote, raising concerns about transparency and the ethical use of AI in professional consultancy work.
Key Facts at a Glance
- Deloitte to refund part of its $440,000 AUD (~$290,000 USD) consultancy fee.
- Report included non-existent academic references and a fabricated Federal Court quote.
- Errors were linked to the use of generative AI tools during the drafting phase.
- The Australian Department of Employment released a corrected version of the report.
- Deloitte claims AI use did not affect the report’s findings or recommendations.
- Refund process is underway, with stricter future guidelines expected for AI use.
What Went Wrong?
In 2024, Deloitte was commissioned by the Department of Employment and Workplace Relations to assess the compliance framework and IT system for welfare-related penalties. However, the final report, released in July, included:
- Fabricated citations referencing individuals who do not exist.
- A false quote supposedly from a Federal Court judgment.
- Over a dozen fictitious references and footnotes.
- Numerous typographical and factual errors.
The inaccuracies were first flagged by Dr Christopher Rudge, a welfare academic, who identified clear signs of AI “hallucinations”—a term used when generative AI creates plausible-sounding but incorrect or fabricated content.
Deloitte’s Response
Deloitte acknowledged the use of generative AI (Azure OpenAI GPT-4o) during the initial drafting process but emphasized:
- AI was only used in the early stages.
- Human experts conducted extensive review and editing.
- The core findings and recommendations remained unaffected.
- The company did not directly blame AI for the errors.
“The matter has been resolved directly with the client,” a Deloitte spokesperson confirmed.
Updated Report Published
Following the discovery of the errors:
- An updated report was published by the Department.
- Fake references were removed.
- Accurate citations were added in their place.
- Methodology disclosure was updated to include AI use.
Wider Implications: AI in Consultancy Under Scrutiny
This incident has triggered public debate and internal government reviews on:
- Transparency in AI use during public-sector consulting.
- Accountability for AI-generated content.
- Financial fairness when clients pay premium fees for human expertise.
The Department has confirmed it may enforce stricter rules for AI use in future contracts to maintain content integrity.
Growing AI Adoption Despite Concerns
Despite the controversy, Deloitte recently entered a deal with Anthropic, granting 500,000 global employees access to the Claude AI chatbot, indicating an industry-wide shift toward integrating AI in professional services.
Conclusion
The Deloitte-Australian government case is one of the first significant AI-related accountability incidents in Australia’s consultancy sector. It underscores the urgent need for clear ethical guidelines, disclosure requirements, and quality control when leveraging artificial intelligence in public-sector projects.
