Insight from Yoshi Soornack & James Garner
A staggering 72% of digital transformation projects fail to meet their goals. Is unchecked AI about to make things even worse?
In the relentless pursuit of efficiency, the professional services industry has embraced artificial intelligence with open arms. But the recent debacle involving Deloitte and a A$440,000 (£230,000) government report serves as a stark, and expensive, reminder that this powerful technology is a double-edged sword. For project delivery professionals, this isn’t just a cautionary tale; it’s a critical lesson in the urgent need for human oversight in an increasingly automated world.
The Anatomy of a High-Stakes Failure
In July 2025, Deloitte’s Australian branch delivered a 237-page report to the Department of Employment and Workplace Relations, intended to help the government clamp down on welfare non-compliance. The report, however, was riddled with errors. A researcher from Sydney University, Chris Rudge, flagged the report as being “full of fabricated references,” a claim that was later confirmed by Deloitte. The firm admitted that some footnotes and references were incorrect and that a generative AI language system, Azure OpenAI, was used in its creation. The embarrassing outcome? Deloitte was forced to issue a partial refund of A$440,000 and quietly publish a revised version of the report.
“I instantaneously knew it was either hallucinated by AI or the world’s best kept secret because I’d never heard of the book and it sounded preposterous.” – Chris Rudge, Sydney University researcher
This wasn’t a simple case of a few typos. The report included a fabricated quote from a federal court judge and references to non-existent academic research papers. As Senator Barbara Pocock of the Australian Greens party scathingly put it, “Deloitte misused AI and used it very inappropriately: misquoted a judge, used references that are non-existent. I mean, the kinds of things that a first-year university student would be in deep trouble for.”
The Illusion of Infallibility
The irony of this situation is that on the very same day the refund was revealed, Deloitte announced a landmark AI enterprise deal with Anthropic, a leading AI safety and research company. This highlights a concerning trend in the industry: a blind faith in the power of AI without a corresponding investment in the critical thinking and verification processes required to use it responsibly. As one of our own at Project Flux commented, “this is just common sense, and while people complain about it on AI, it’s no different than somebody using Excel or the internet and not checking things. It’s ridiculous to use this as the basis to say all AI is doomed and that this is all just a case of ‘don’t be stupid.’”
This is not an isolated incident. The Chicago Sun-Times was forced to admit it had run an AI-generated list of books with hallucinated titles, and even Anthropic’s own lawyers have apologized for using an AI-generated citation in a legal dispute. The problem of AI “hallucinations” – where the model generates false information – is a known issue, yet it seems many are willing to overlook it in the race to adopt the latest technology.
The Project Manager’s AI Mandate
For project managers, the Deloitte case is a critical wake-up call. The pressure to deliver projects faster and cheaper is immense, and AI offers a tempting solution. But as this incident demonstrates, the risks of unchecked AI are equally immense. The reputational damage to Deloitte is significant, and the financial cost is a stark reminder that cutting corners with AI can have serious consequences.
So, what can project managers do to avoid a similar fate? The answer lies in a renewed focus on human oversight and critical thinking. Here are three key takeaways:
- Verify, Then Trust: Never blindly trust the output of an AI model. Treat it as a starting point, not a finished product. Every fact, every figure, every citation must be verified by a human expert.
- Invest in AI Literacy: Your team needs to understand the limitations of AI as well as its capabilities. Invest in training that teaches them how to spot AI-generated errors and how to use the technology responsibly.
- Build in Quality Assurance: Your project plan must include a robust quality assurance process that specifically addresses the risks of AI. This should include multiple rounds of review by subject matter experts and a final sign-off by a senior team member.
Don’t Be the Next Cautionary Tale
The Deloitte AI blunder is a warning shot for the entire professional services industry. As project managers, we are on the front lines of this technological revolution. It is our responsibility to ensure that we are using AI to enhance our work, not to undermine it. The future of our projects, and our profession, depends on it.
Ready to future-proof your project management skills? Subscribe to Project Flux for the latest insights on how to navigate the age of AI and deliver projects that succeed.
References:
- Deloitte was caught using AI in $290,000 report to help the Australian government crack down on welfare after a researcher flagged hallucinations
- Deloitte goes all in on AI — despite having to issue a hefty refund for use of AI
- AI on Trial: Legal Models Hallucinate in 1 out of 6 (or More) Benchmarking Queries
- Big 4 struggle with AI adoption, while boutique firms win
- AI Is Changing the Structure of Consulting Firms
- Why AI auditing needs human oversight
