The department continues to restrict student access to generative AI applications on devices within department networks. As part of its commitment to duty of care for all its students, the decision to restrict access is due to the lack of reliable safeguards preventing these tools from exposing students to potentially explicit and harmful content. While department staff have access to generative AI tools, the following guidelines are recommended for safe, ethical usage.

These guidelines are intended for informative purposes only. The use of individual free-to-use generative AI tools is neither endorsed nor recommended by the department. While not endorsed, staff still have access to free-to-use generative AI tools. This page provides high-level advice, and more comprehensive advice has been updated on the artificial intelligence in education intranet page to support staff in safe, ethical usage.

Safeguard personal information

When using generative AI tools for teaching and learning, it is essential to manage personal information effectively and ensure proper de-identification processes are in place.

What is personal information?

Personal information is any information that can be used to identify an individual directly or indirectly. It could be a student’s name, address, class, school, family details, fingerprints or a combination of information from which a student or other individual can be identified. The information can be in recorded in paper files, electronic records, video recordings and photographs.

For more information, visit Managing personal and health information.

De-identification of personal information

De-identification involves removing, anonymising, or masking personal information so that it can no longer be used to identify an individual. This process is crucial to protect privacy and comply with data protection regulations.

The following techniques will help you safeguard sensitive information and comply with data protection regulations while maintaining the utility of the data for analysis and learning purposes.

• Data anonymisation: Replace or alter personal information in the text using information that cannot be linked to the individual.
• Data masking: Replace personal information in the text, such as email addresses or phone numbers, with fictional data that retains the same format.
• Include only relevant data: Choose only information that is relevant.

Verify and edit AI generated content for accuracy and suitability

The department supports the NSW mandatory ethical principles for the use of AI, and advises all DoE staff to verify and edit AI generated content to ensure accuracy and suitability of the output. Teachers should be carefully assessing if the AI generated content supports the first two teaching Standards.

1. Know students and how they learn
2. Know the content and how to teach it.

There are currently no tools that can reliably detect AI generated content and their output should not be used as evidence of student use of generative AI.

Stay up to date with training

The department requires all school staff to stay-up to date with mandatory training, such as cybersecurity, data breach response and child protection training. The department advocates for protection of personal information, including sensitive information.

Use effective prompting

Writing a clear and effective prompt can greatly improve the quality of output from generative AI tools, while also reducing errors or irrelevant responses. Provide clear roles and examples when creating a prompt, which involves defining a character and role for the AI tool, giving a clear task, providing examples and additional guidelines, specifying the type of output required, and including any extra phrases to improve the prompt. These steps help the AI model understand the task more accurately and produce results that meet the required specifications. See this article for an overview, or for training in prompting, see How to research and write using generative AI tools.