A Dentons employment partner warns AI-generated meeting notes and prompts are increasingly discoverable in unfair dismissal cases
As Australian tribunals grapple with a surge in generative AI use, Nick Linke, partner and Australian head of employment and safety at Dentons, is warning HR practitioners that the AI note-takers and chatbots they use every day could end up as evidence against them.
Speaking to HRD, Linke said the growing use of AI note-takers in internal HR and management meetings – including discussions about redundancies and terminations – is creating a real-time, complete transcript of decision-making that did not exist a decade ago.
"If you end up with an unfair dismissal claim or a general protections claim, and those transcripts are relevant, they might then be discoverable in the proceedings," he said.
A transcript problem HR didn't have before
Under Australia's discovery rules, documents relevant to a dispute – including electronic records – can be ordered to be produced to the other side and the Fair Work Commission (FWC) during a case.
Linke said that shift changes the risk calculation for HR teams using AI tools to summarise sensitive conversations.
"Someone may take notes or may not take notes deliberately, and then they'd move on," he said, contrasting today's practice with the past. "Now you've got this real-time complete transcript."
That matters most when a manager makes an offhand comment during a recorded meeting – for example, referencing an employee's sick leave history – that could support a general protections claim.
Linke said comments like that can become "dynamite" once captured in a transcript that ends up before the Commission.
Deleting the record can look worse than keeping it
Some HR teams have responded by having AI note-takers summarise meetings and then deleting the underlying transcript.
Linke cautioned this creates its own exposure. Destroying a transcript that could have assisted a tribunal in reaching a decision, he said, may expose an employer to criticism for destroying evidence – even where deletion is a deliberate, standing practice.
He added a further complication: "Even when you think something's deleted, it often isn't," pointing to the growing sophistication of forensic computer specialists able to recover material that has ostensibly been removed.
Linke's advice is for HR to treat every AI-assisted meeting record as though it will eventually be read in a courtroom.
"If they use AI as a tool to do something, they need to make sure that however they're instructing AI is something that they'd be happy to have produced in court," he said.
The legal profession is already paying the price
Linke's warning lands as Australian courts and tribunals confront a wave of AI-related failures among legal practitioners and litigants themselves.
In September 2025, a Victorian lawyer known as "Mr Dayal" became the first in Australia to be sanctioned over AI use, after the Victorian Legal Services Board and Commissioner stripped him of his ability to practise as a principal following unverified, AI-fabricated citations filed in the Federal Circuit and Family Court.
Around the same time, senior Melbourne barrister Rishi Nathwani apologised to the Supreme Court of Victoria after AI-generated fake quotes and non-existent judgments were submitted in a murder case, delaying proceedings by 24 hours, according to reporting at the time.
The FWC has faced a parallel problem from the other direction. In February 2026, FWC president Justice Adam Hatcher told the Victorian Bar Association that the Commission's overall workload had grown more than 70 per cent in three years, a rise he attributed largely to litigants using generative AI to prepare unfair dismissal and general protections applications, some citing legal authorities that do not exist.
This is why the Commission moved to formalise disclosure requirements for AI-assisted filings earlier this year.
Linke said he is not aware of an Australian HR practitioner being caught out in a case specifically because of AI use – but expects that to change.
"I think it's only a matter of time," he said, adding that published tribunal decisions are likely to feature HR-related AI missteps "in the not-too-distant future."
Balancing productivity against exposure
None of this means HR should avoid AI altogether, Linke said. The technology brings genuine productivity gains, and its use in HR functions is only expanding – a trend reshaping how HR teams and workforce planners operate more broadly.
The task for HR leaders, he said, is balancing that productivity against the reality that AI-generated material – prompts, summaries, and transcripts alike – is now part of the evidentiary record in employment disputes, whether or not it was ever intended to be.
That balancing act sits alongside other legislative uncertainty HR teams are navigating. For now, his message on AI is straightforward: use it carefully, verify everything it produces, and assume that whatever goes in – and comes out – could one day be read aloud in a hearing room.
