Graeme KerrCONSULTING

AI & Algorithmic Transparency Statement

Last updated: June 2026

We help Scottish businesses use AI in a practical, honest way. This statement explains how we use AI ourselves — in client work and on this website — so you can make informed decisions before engaging us or sharing data with us.

It complements our Privacy Policy, Cookie Policy, and Terms of Use.

1. AI on this website

  • This website does not currently use chatbots, generative AI features, or automated decision-making affecting visitors.
  • We do not profile visitors. Analytics (if you opt in) is aggregate only — see our Cookie Policy.
  • Some marketing copy and insights articles on the site are drafted with AI assistance and then reviewed, edited, and approved by a human before publication.

2. How we use AI in client work

We use AI tools day-to-day to accelerate research, drafting, code generation, and analysis. The specific tools vary by engagement but typically include large language models (LLMs) from established providers (e.g. OpenAI, Anthropic, Google) and developer tools that wrap them.

Our default principles when using AI on a client engagement:

  • Human in the loop. A human reviews and is accountable for every AI-generated output we deliver. We do not ship raw model output without review.
  • Fit for purpose. We choose models and architectures appropriate to the task and risk level. We are open with clients about what is automated and what is human-produced.
  • Honest about limitations. AI systems can be wrong, biased, or out of date. We surface these limitations and design controls around them rather than hide them.
  • No covert use. We tell clients when AI is materially involved in producing a deliverable.

3. How we handle client data with AI tools

  • We do not feed client confidential information into consumer AI tools.
  • For client work we use business or enterprise tiers of AI providers where inputs and outputs are not used to train the provider's models.
  • We minimise the data shared with any model — we use only what's necessary for the task, and we anonymise or redact personal data where possible.
  • For sensitive engagements we can agree additional controls in writing, including approved tool lists, on-premise / private model options, retention limits, and zero-retention API configurations where the provider supports them.
  • Specific data handling, sub-processors, and international transfers for a given engagement are set out in the Data Processing Addendum to your contract.

4. Automated decision-making

We do not use AI to make automated decisions that have legal or similarly significant effects on individuals — either on this website or, by default, in the systems we build for clients. Where a client engagement involves any kind of automated decision-making, we design it with human review, explainability, and an appeal/override path, and we document this with the client.

5. Bias, accuracy, and intellectual property

  • Accuracy. Generative AI can produce plausible but incorrect information ("hallucinations"). We treat AI output as a draft to be verified, not a source of truth.
  • Bias. Models reflect biases in their training data. For any AI-assisted output that could affect people (e.g. hiring, lending, eligibility), we apply additional review and discuss bias risk explicitly with the client.
  • Intellectual property. We are mindful of IP and copyright risk in model outputs. We do not knowingly pass off third-party IP as client-owned, and we discuss IP ownership and licensing of AI-assisted deliverables in the engagement contract.

6. Regulatory context

We track relevant UK and EU developments, including the UK's pro-innovation AI regulatory framework, the EU AI Act, and ICO guidance on AI and data protection. Where an engagement involves systems that may fall within scope of these regimes, we flag this to the client and design accordingly.

7. Reporting concerns

If you believe an AI system we have built or advised on has produced a harmful, biased, or incorrect result, please contact us at gk@graemekerr.co.uk. We will investigate, respond, and where appropriate work with the client to remediate.

8. Changes to this statement

We may update this statement as our practices, tools, and the regulatory landscape evolve. The "Last updated" date above reflects the latest revision.