AI policy

Our approach

Gro's analysis comes from quantitative models, not AI. The models are deterministic: the same inputs always give the same result. AI is used only in the places listed below, and the adviser always decides.

Where Gro uses AI

• Plain-English explanations of portfolio results, so advisers can read and share them quickly.
• Scoring the sentiment of news and company filings, with a financial language model.
• AI prompt blocks that a firm chooses to add to its own strategies.

What AI never does

• It never gives financial advice.
• It never places a trade on its own.
• It never replaces the quantitative models behind Gro's analysis.
• It never makes a decision about a client without an adviser reviewing it.

How AI use is controlled

• Every AI output is checked against a fixed format before Gro uses it. An output that fails the check is discarded.
• AI prompts never receive credentials or client identifiers, such as names or account references.
• Firms choose whether to use AI prompt blocks at all.
• Each use of AI is recorded in the audit trail.

Your data

Gro does not use firms' or clients' data to train AI models.

AI requests run on Google Cloud in the UK (London). Google's Service Specific Terms (Training Restriction) state that Google will not use customer data to train or fine-tune any AI models without the customer's prior permission or instruction. Google explains this in its data governance guide.

The financial language model that scores sentiment runs on Gro's own infrastructure.

Oversight

We review how AI performs in Gro, test changes before release and act on reported problems. This policy follows the UK's principles for AI: safety, transparency, fairness, accountability and the ability to challenge outcomes.

Contact

To ask about AI in Gro or report a concern, use the contact form on our home page or write to us at Cell Block Studios, Portsmouth Historic Dockyard, College Road, HM Naval Base, Portsmouth PO1 3LJ.