Sovereign AI Toolkit
Sovereign: AI you run on your own systems, in line with your obligations on data, copyright, and accountability.
Open model: an AI model that is open source. You can download it and run it on your own computer. Closed products like ChatGPT and Claude cannot. Examples: Llama, Mistral, Qwen.
This toolkit recommends open models, run on your own systems.
| AI Readiness Area | What sovereign looks like | Where to start |
|---|---|---|
| Leadership & governance | Leadership can name risks, evaluate them, and hold a position | Leadership position |
| GDPR, data protection & confidentiality | Sensitive material processed locally | Block A |
| Approved tools & access | A sanctioned set of AI tools staff can actually use | Block B |
| Staff use, skills & confidence | A safe place for staff to use AI | Block A and Block B |
| Workflow & productivity opportunities | AI embedded in real work | Block B |
| Knowledge, archive & reusable assets | AI that works with your own documents | Block C |
| Creative rights, IP & accountability | Creative material handled locally, with audit trails | Block A |
The next step. Identify which areas are most pressing for your organisation.