Environmental impact (optional)

Overview

Training and running AI uses energy, water and hardware. This is an optional topic for most organisations, but some customers and regulators ask, and simple choices such as model size and hosting location can reduce impact.

Key risks

• High energy use for large models
• Wasteful always-on automation
• No visibility from vendors

What to do now

• Ask vendors what they can share.
• Use smaller models where they are enough.
• Avoid unnecessary calls and loops.
• Include it in procurement criteria if it matters to you.

Questions to ask

• Do we know roughly what our AI workloads consume?
• Could a smaller model do the job?
• Does the provider report on energy and emissions?
• Do customers or regulators ask us for this information?
• Who owns the question internally?

Related tools

• Vendor due diligence

Region rules: EU · US · UK · India · Global standards.

Reviewed 3 Oct 2026. Sources: European Commission: AI Act overview.