AI & Data Protection Statement

How We Use AI — and How Your Data Stays Protected

Lighthouse Advisory Ltd. · Last updated 12 September 2026

Lighthouse Advisory uses artificial intelligence selectively — to research faster, draft better and check our own work more rigorously. We never use it at the expense of your confidentiality. This statement explains, in plain terms, what we do and don’t do with AI, and the protections that apply whenever your information is involved. It sits alongside our Privacy Notice under Bermuda’s Personal Information Protection Act 2016 (PIPA).

What we use AI for — and what we don’t

We may use AI tools to: accelerate research on regulation and market practice; draft and refine documents; summarise and structure material; and quality-check calculations and text.

We do not use AI to: make judgments or form the advice we give you — professional conclusions are always reached and verified by a qualified professional; nor do we ever place your data in free or consumer-grade AI services.

Our six commitments on your data

1. Your data is never used to train AI models.

We only use AI tools under commercial (business-grade) terms that contractually prohibit the provider from training its models on anything we submit — our inputs and the outputs alike. This is a term of contract with the provider, not just a settings switch.

2. No long-term storage by AI providers.

The commercial terms we require mean data submitted to an AI tool is deleted from the provider’s systems within a defined short period — currently 30 days or less — and stricter zero-retention arrangements are available where an engagement warrants them. Your files continue to live in our own secured document environment, not in AI services.

3. Business-grade tools only, formally approved.

Every AI tool we use must pass a written vendor assessment — no-training commitment, bounded retention, data processing agreement, encryption, and strong account security — and is recorded in our PIPA vendor register before it touches any client material. Free and personal AI accounts are prohibited under our internal AI Use Policy.

4. Minimum data, anonymised first.

Only what is needed for the task enters an AI tool. Where the work involves personal information — for example claim-level data in reserving or expert-review engagements — our default is to anonymise or aggregate it before any AI processing, and sensitive personal information (such as health data) is not entered in identifiable form.

5. A professional reviews everything.

AI output is treated as a draft, never a deliverable. Every figure, citation and conclusion is verified by the responsible professional, who remains fully accountable for the work under applicable actuarial professional standards.

The same applies to what we publish. AI tools are used in the research and drafting of our articles, posts and other published analysis; every piece is reviewed and owned by its author, reflects views we hold and analysis we have actually done, and is checked against primary sources before it appears under the Lighthouse name. This statement is our standing disclosure of that practice.

6. Bermuda PIPA compliance, end to end.

Lighthouse has a designated privacy officer, assesses every overseas transfer of personal information (PIPA s.15) before an AI provider is approved, and operates a breach-response plan: in the unlikely event of an incident likely to adversely affect an individual, we would notify the Privacy Commissioner and affected individuals without undue delay.

Your choice

AI use is our working method, not a condition of engagement. If you prefer that your confidential information not be processed with AI assistance, tell us — in your engagement letter or at any time in writing — and we will deliver the work without it. Your instruction is recorded and honoured for the remainder of the engagement.

Questions

We are glad to walk through our AI Use Policy, our vendor assessments, or the specific tools proposed for your engagement.