Using AI in applications

Using AI is completely optional and not expected. Our guidance is here to support those who wish to use these tools, but success does not depend on them.

Your application is your chance to tell us about yourself and show us what you can do. AI can be helpful when used appropriately, but it should only be used to amplify your talent and not distort it. We want the real you to shine through. Please note that we don’t use AI to make our hiring decisions; your application will be read and assessed by a human.

And please remember, you’re applying for a job in national security, where protecting personal and organisational data is essential. If you choose to use AI tools, you should do so securely and responsibly. Do make sure you think carefully about how you might use it.  To help, we’ve put together some practical advice on how, when and where AI can support your application.

How can I use AI securely?

When applying for a job in national security, we ask that you take active steps to protect both your privacy and that of HMGCC when using AI tools. Many AI platforms store, reuse or share the data you input. This means that entering personal or organisation information such as your name, address, employment history, or references to HMGCC could unintentionally expose sensitive details. 

To reduce this risk:

• Avoid entering the name of HMGCC and / or any personal data into AI platforms.

• Delete any data you’ve entered once you’ve finished using the tool.

• Review the platform’s privacy settings and terms of use.

• Seek advice from us if you’re unsure about the safety or appropriateness of a particular tool.

When do we ask you not to use AI?

At HMGCC, we place great value on honesty, truthfulness and integrity. These qualities are at the heart of who we are and what we look for in candidates. AI can be a valuable tool to use in advance of application and interview stages. However, there are parts of the process where the use of AI tools is strictly prohibited. Please note we may need to ask you to retake a test or interview under controlled conditions if we suspect AI has been used in contravention of this guidance. 

AI use is prohibited for the following stages (please note your assessment process may not involve all of these):

• Online tests 

• Online, telephone or face to face interviews

• Automated video interviews

• Any data analysis or written exercises

• Interactive assessment centres

Remember: we’re looking for authenticity. As such, AI should only support (and not replace) your personal experiences and shouldn’t be used in a way that misrepresents who you are.  Relying too heavily on AI at any point in the application can create gaps between what’s presented and your actual experience, which may become evident as the process continues or once you’re in the role.

Dos and Don’ts

Do:

 • Feel free to use AI to review your phrasing or formatting, or to help prepare answers to potential interview questions in advance of the real thing.

• Be prepared to discuss anything you put in your application. 

• Review the AI output to ensure it reflects the real you and hasn’t introduced incorrect facts or bias.

• Use it as a preparation tool to help strengthen your answers and examples (making sure to protect your privacy).

• Get in touch if you are uncertain how you might use AI responsibly at any point during the hiring process.

Don’t:

• Allow AI to complete your application on your behalf. If your answers don’t match the real you, this is likely to become apparent later in the process.

• Use AI tools to help you during live interviews or assessments.

• Input private information about yourself or HMGCC into AI platforms.

• Use AI-generated answers directly in your application without reviewing and editing them to reflect the real you.
  

You can find detailed guidance online on how AI works and how to write AI prompts effectively, as well as information on AI’s limitations, weaknesses and risks.

For example, see the National Cyber Security Centre website: - ChatGPT and large language models: what's the risk?