How IMBER uses AI

IMBER uses artificial intelligence to accelerate the development of our platform. We aim to be transparent with how we use this rapidly developing technology, and are committed to using it responsibly.

Over the course of the development of our platform, we have used two frontier AI models to assist with the coding of the data conversion systems, API, and front-end UI. We use frontier AI models as they have higher accuracy, faster responses, and generally lower rates of hallucination. This allows us to develop much faster.

In the first week of the project, we used OpenAI’s ChatGPT 6.0 Sol. In November 2026, we moved our development to Anthropic’s Claude Opus 5.5. We made this change due to Claude Opus 5.5’s strong performance on tasks our platform depends on, and Anthropic’s commitment to transparency and safety.

We’re also acutely aware of the challenges posed by artificial intelligence, including its environmental impact. We believe that the placement of data centres and the regulation of AI safety are crucial issues that research and policy-making must address. At the same time, we believe AI can be an enabler for climate-positive technologies like IMBER, and we take a pragmatic approach to using it.

We are setting out a series of principles to ensure that we are as transparent as possible with our use of AI. Most of these have been followed since our founding, and all will be followed from 6 November 2026:

1. Human oversight. A member of our team tests all AI-assisted code before it is deployed. IMBER remains fully responsible for how our platform behaves.
2. Data integrity. AI is used to help write the systems that process weather data. It does not generate, alter or interpret the data itself, unless a member of our team has decided to use it this way and checks the results.
3. Privacy. We never enter personal user data, passwords or API keys into AI tools. Where possible, we opt out of our inputs being used to train AI models.
4. Security. All AI-assisted code is checked for security risks before it is deployed, and is not used where we cannot confirm it is secure.
5. Labelling. Blog posts written with the help of AI will be clearly labelled as such.
6. Keeping this up to date. We will update this page whenever the AI models we use change, and review these principles at least once a year.
7. Feedback. If you have any questions or concerns about how we use AI, you can contact us at benjaminraisanen@hotmail.co.uk.

Last updated: 6 November 2026