Students at two Sydney schools were required to resit their NAPLAN writing tests after Apple's AI-powered predictive text feature was inadvertently enabled during the assessments, potentially providing unfair assistance with spelling, word completion, and text generation.
In 2025, Year 5 students at Waverley College and Kambala school in Sydney were forced to retake their NAPLAN writing examinations due to a technical issue where Apple's predictive text feature remained enabled during testing. The Apple feature uses artificial intelligence to assist with spelling, word completion, and creates text responses suited to the user's writing style. A teacher at Kambala and students at Waverley discovered the issue during the writing component of the test. Both schools immediately reported the problem to the NSW Education Standards Authority (NESA) and the Australian Curriculum and Reporting Authority (ACARA). Students were required to sit the test again with a new prompt, though Waverley College experienced delays due to missing student profile codes needed to access the rescheduled test. The incident was limited to Year 5 students at these schools, with Year 3 students unaffected as they handwrite their responses. ACARA stated that over 880,000 online writing tests had been successfully submitted nationwide in the first four days of testing, indicating this was not a widespread issue. Schools had been provided instructions to disable predictive text features before testing and were reminded again on the second day of testing.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
Human
Due to a decision or action made by humans
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed