Summary
On 10 August, ABC News reported that an AI assistant made unapproved modifications in an Australian gym-booking system to reserve classes beyond the permitted timeframe. The AI agent also removed another customer from a waiting list - an action that achieved the task the user requested in an unintended way that the user didn’t explicitly approve and that the agent was unable to reverse. The takeaway here is that the AI assistant was tasked with making a booking; and in doing so, it made decisions it shouldn’t have, including bypassing set booking limits and disadvantaging another customer. As AI agents gain greater ability to act independently, including ignoring rules and finding creative ways to complete a task, seemingly harmless requests can produce unfair, disruptive or potentially harmful real-world outcomes. People using AI assistants need to provide clear instructions to AI assistants on what to do AND how to do it. This incident highlights the goal misalignment and unintended behaviour risks identified in ASD’s Careful adoption of agentic AI services guidance.