Cookies on the Defra digital service manual

We use some essential cookies to make this service work.

We'd also like to use analytics cookies so we can understand how you use the service and make improvements.

View cookies
Skip to main content
Department for Environment, Food and Rural Affairs: AI digital toolkit
AI digital toolkit

Deliver with AI

This guidance is for digital delivery teams building with AI on Defra services. It is run by the AI Capability and Enablement team (AICE).

It applies the AI Playbook for the UK Government to the way Defra delivers, so the rules here line up with the 10 principles all government teams follow.

Not a digital delivery team? Find where your work fits

Several teams across Defra work on AI. If your work is closer to one of these areas, start there instead.

  • AI training for all staff. Civil Service Learning courses and safe-use guidance. Complete this before using AI on Defra work. See the AI training and guidance page on SharePoint.
  • AI policy and governance. Defra-wide policy on what is allowed and what needs sign-off. See the AI governance page on SharePoint.
  • Core Delivery Platform (CDP) and Amazon Bedrock. Approved models, guardrails and cost controls for the Defra Bedrock integration. See the CDP documentation.
  • AI in arm's length bodies. Environment Agency, Natural England, APHA, RPA, MMO and others run their own AI work. Check your organisation's own guidance first.

Still not sure? Email AICE at AICapabilityAndEnablement@defra.gov.uk and we will point you to the right place.

Choose tools and use data

  • Choosing a tool. Check a tool is approved, turn on privacy settings, spot the AI hidden in everyday tools.
  • Using data with AI. What you can put where, by data classification.
  • Keeping data safe. Classification, where your data goes, and removing personal data.
  • Working with AI agents. What teams have learned about agent identity, tools, evaluation and observability.

Use AI responsibly

The responsibility side of using AI at Defra. Use these alongside the rules above, not after.

  • Security. Keeping AI-generated code, secrets and your environment secure.
  • Ethics. Accountability, bias, transparency and public trust.
  • Sustainability. Using AI in proportion to the task.
  • Report an AI incident. What to do if data has been shared with AI inappropriately.

Where else to look

Ask the AI Capability and Enablement team

Get hands-on help with a project, or a position on something this guidance does not yet answer.