Why this matters

Invest AI surfaces BUY / HOLD / SELL style signals and confidence scores. In many jurisdictions, anything that looks like personalized investment advice can drag you into regulated territory unless you are very clear about what you are — and what you are not.

We are not a Registered Investment Advisor (RIA). The product is framed as an educational tool: explore market data with AI, do your own research, talk to a licensed professional before acting.

This post is not legal advice. I am not a lawyer. Before any paid or high-traffic commercial launch, a securities lawyer should review the exact copy. For a free public beta, we still wanted a defensible, industry-standard pattern that is hard for a reasonable user to miss.

The five layers

We stack five independent surfaces so the message survives UI changes, scrapers, and future API clients.

flowchart TD
    L1["Layer 1 · Persistent footer<br/>every page"]
    L2["Layer 2 · First-visit modal<br/>must accept"]
    L3["Layer 3 · Inline AI badge<br/>next to recommendations"]
    L4["Layer 4 · Terms + Privacy<br/>full pages"]
    L5["Layer 5 · API disclaimer field<br/>every JSON response"]

    L1 --> L4
    L2 --> L4
    L3 --> L4
    L5 --> L4

1. Footer — One short line plus links to Terms, Privacy, Help, and contact. Always visible.
2. First-visit modal — Bulleted summary, checkbox acknowledging Terms + Privacy, no dismiss without acceptance. Versioned localStorage key so we can force re-accept if copy changes materially.
3. Inline badge — Next to recommendation-heavy UI: “AI · Not Advice · You Decide” with a tooltip.
4. Terms & Privacy — Full pages at /terms and /privacy.
5. API envelope — Every /api/v1/analyze response includes a disclaimer string. Third-party clients and scrapers still see the legal framing even if they strip the HTML.

Single source of truth (almost)

Frontend legal copy is centralized in lib/legal/disclaimers.ts. The backend mirrors the same substance in ANALYSIS_DISCLAIMER so API responses stay aligned. A future improvement is one shared artifact (JSON/YAML) imported by both stacks.

What we deliberately did not do

• Become an RIA — Wrong product stage; high cost and ongoing compliance.
• Footer only — Too easy to miss; regulators care about prominence.
• Hope nobody complains — Not a strategy.

Takeaway

If you ship an AI product that touches money decisions, treat disclaimers as product design, not a legal appendix. Five thin layers beat one thick paragraph nobody reads.

Further reading: ADR-005 in the Invest AI repo (docs/adr/005-legal-disclaimers-v1.md).