Legal: review AI output framing and UK defamation exposure before monetisation

[mahsa7haft]

Background

PaidUp uses Claude to surface "conflicts of interest" and "factional leanings" about named, real UK politicians. This is fine as a free tool presenting AI interpretation of publicly declared data — but the legal stakes change if the product is monetised.

Actions needed before charging money

• Ensure all AI-generated output is clearly and prominently labelled as AI interpretation of declared register data, not verified fact — not just in small print but visually obvious in the UI
• Review all four prompt outputs (summary, investigative, donor profiles, gaps) for language that could be read as stating fact rather than interpretation
• Get a quick legal view on UK defamation exposure — UK law is stricter than US; the "responsible publication" defence and honest opinion defence are relevant here
• Consider adding a consistent disclaimer header to every AI analysis output in the drawer
• Review whether the "investigative journalist" prompt framing in particular needs softening

Notes

• Data source is the official UK Parliament Register of Members' Financial Interests — publicly declared, so the underlying data is factual
• The legal risk is in the AI's interpretation of that data (conclusions about conflicts, leanings, affiliations)
• This is not urgent for a free tool but is a blocker before any paid tier