Source: ABC News

The Fair Work Commission in late July ordered an employee and their employer to cease using artificial intelligence tools for all future correspondence after ruling that the employee had been bullied. The dispute concerned a caretaker for a Queensland body corporate who submitted that the workplace repeatedly refused to clarify duties, gave confusing and shifting directions, and raised several other allegations; Commissioner Sarah McKinnon found the caretaker had been bullied and examined emails from both sides.

Commissioner McKinnon described the caretaker’s correspondence as “lengthy, wide-ranging, replete with generalisations, repetitive, and often couched in accusatory language,” and noted the emails appeared to have been prepared using an AI tool. The commissioner said the workplace’s decision to “disengage and dig in” meant it did not meaningfully address many of the caretaker’s concerns, and to prevent further bullying ordered that all future communications be accurate, respectful, brief and that “there is to be no use of artificial intelligence tools in the preparation of correspondence between the parties,” with the order to take effect immediately. Griffith University’s Mai Nguyen told the article that using AI to improve professionalism and grammar is now common but warned the wrong prompt can exaggerate facts and produce an inappropriate tone. QUT digital communication professor Daniel Angus said people are tempted to use AI to make text more forceful or persuasive and urged normalising disclosure of AI use while acknowledging disclosure is “a really tricky topic”; he also warned about a future where workplaces may consist of “a bunch of AI agents talking to each other.” University of Melbourne law research fellow Guzyal Hill noted few organisations currently have AI policies and stated “Australia does not have an AI Act, so each organisation has to develop its own AI Strategy/AI Policy.” The AI Governance Chair at HWLE Lawyers said skilful, legally compliant use of AI for emails may cause little harm but should be discouraged when private or confidential information is entered into AI generators; the piece also reports that some academics find humans “notoriously” bad at detecting AI-written text and advise double-checking AI-assisted messages for clarity and intent.