NOTE: This article was originally an op-ed on the risks of unchecked AI writing in Australian workplaces. We’ve lightly edited it below to share it with you on our blog.
—————
The most common question I hear from professionals in my training rooms is not whether they should use AI for their business writing. It’s: How can I use it without anyone finding out?
That tells you everything you need to know about the state of workplace writing in the age of AI.
According to EY research, 68% of Australians are already using AI at work, but most are doing it without guidance or training – and apparently, without much confidence that the results are any good. They’re right to worry.
AI-generated writing may look professional on the surface, but spend a few minutes reading carefully and the gaps become clear. No precision, no real point of view – and no sense of a real person behind the words. Spend a few more… and even more troublesome tells emerge:
statistics that can’t be traced and sources that don’t exist.
This is the everyday cost of AI business writing without oversight. Employees are using it to produce work they lack the skills to evaluate and the results are landing in front of clients and stakeholders who notice.
The failures that make headlines are starker.
Last October, Deloitte submitted a 237-page report to the federal government containing fabricated citations and a quote falsely attributed to a Federal Court judge. The firm refunded part of its AU$440,000 fee, but the reputational cost is harder to quantify.
A couple of months later, a paralegal lost their job after AI-generated documents went unchecked by a supervising solicitor and two barristers, resulting in a $10,000 costs order against the firm.
And then again, in the Supreme Court of Victoria, a King’s Counsel apologised after AI-generated submissions in a murder trial were found to contain fictitious case citations and quotes. The judge was direct: “It is not acceptable for artificial intelligence to be used unless the product of that use is independently and thoroughly verified.”
These cases are the sharp end of a much broader problem.
For every fabricated reference that gets caught, there are emails that bury the point in corporate word-salad and proposals a discerning reader would find unconvincing in minutes.
And it’s costing workplaces far more than they realise. According to a 2026 global study by Workday, nearly 40 per cent of employees say that the time saved by AI is cancelled out by rework, including rewriting content and verifying unreliable outputs.
Rather than reducing workload, AI is redistributing it to the people with enough experience to recognise what’s wrong.
This data matches what I see in organisations every day. Managers are spending valuable time rewriting their team’s emails, reports and proposals because they can’t afford for poor communication to leave the building. So the work still gets done – but at a much higher cost to the business.
At the heart of the issue is a capability gap: organisations have handed employees a powerful writing tool without teaching them how to judge the quality of what it produces.
This has been compounded by the mistaken belief that AI reduces the need for writing skills. In reality, it raises the bar. The easier it becomes to generate content, the more important it is to know what good looks like and how to refine the output.
In response, many organisations are pouring their training budgets into AI literacy. That’s not a mistake. Employees do need to understand how these tools work. But AI literacy is not the complete solution.
The missing piece is targeted investment in business writing skills – teaching people how to write with clarity and conviction and how to fact-check with genuine rigour. The legal professionals who submitted fabricated case citations in the murder trial had checked some references and assumed the rest would be fine. That assumption is precisely what good writing training exists to prevent.
Skilful writing in the workplace has always mattered. AI hasn’t diminished its importance. If anything, it has exposed how valuable it really is.
Organisations that fail to recognise this will pay the price – in costly rework, reputational damage and apologies to judges.
—————
What this means for leaders managing teams that use AI to write
- Update your quality assurance process. AI-assisted writing should be reviewed for accuracy, context and strategic judgement – not simply whether it reads well.
- Teach teams on what good business writing looks like. People can’t judge AI output if they can’t spot weak writing when they see it.
- Create AI writing guidelines. Set clear expectations about when AI should be used, when it shouldn’t and how every draft should be reviewed.
- Build critical thinking into the process. Fact-checking, questioning assumptions and verifying sources should be expected every time.