Guide · CompanionGuard · Published 14 November 2026 · Reviewed 14 November 2026 · By the Ingenious AI team
Embedding the Disclosure Snippet: The Free Banner That Carries Disclosure and 988
The short answer
The disclosure snippet is the free kit's embeddable piece of JavaScript: placed on the bot's own pages, it shows the AI disclosure banner with the crisis line built in ('You are talking with {bot}, an AI companion — not a human. If you are in crisis, call or text 988…'), and with minor mode on it also delivers break reminders at least every 3 hours for known minors. It is the mechanism that satisfies the disclosure duty where the user actually is — on your pages, from the first interaction.
What is the disclosure snippet?
The disclosure snippet is embeddable JS issued with each bot's free CompanionGuard kit. It renders the disclosure banner on the operator's own site or app surface — naming the bot, stating it is an AI and not a human, and carrying the crisis line — and it is the delivery vehicle for minor-mode break reminders. Each bot also gets a hosted demo/embed surface (its /demo page and /embed script) from the same kit.
Create the free safety kit Next guide All guides
CompanionGuard is built by Ingenious AI. A compliance toolkit — not legal advice and not a clinical service. If you are in crisis right now: US — call or text 988; UK & Ireland — Samaritans 116 123.
Why disclosure must live on your pages, not ours
A disclosure that lives only on the vendor's page discloses nothing to your user. The duties in the packs attach to the interaction — first interaction, on your surface, in your product's flow — so the banner must render where the conversation happens. The snippet is the delivery mechanism: one embed, issued per bot at kit creation, rendering the pack's banner wording with your bot's name substituted into '{bot}'.
Embedding also solves the partner problem: when your bot is iframed, white-labelled or demoed on someone else's page, the disclosure travels inside the embed rather than depending on the host page's copy. The transparency guide tells the story of a partner embed that cropped the disclosure out — the failure this design prevents.
What the banner says, and why each clause earns its place
The California and Colorado banners share a skeleton: 'You are talking with {bot}, an AI companion — not a human. If you are in crisis, call or text 988 (Suicide and Crisis Lifeline) any time, free and confidential.' Clause by clause: the bot's name (the disclosure attaches to this conversation), 'an AI companion — not a human' (the fact, in words that cannot be misread), and the crisis line with its access qualifiers (free, confidential, any time). The state minor variants append their reminder promise — California's explicit 3-hour wording, Colorado's session check-in.
Resist editing the skeleton for brand voice. The banner is a legal-ethical instrument wearing a UI component's clothes; its wording has been chosen for clarity under stress, not charm. Your brand voice has the whole rest of the page to live in. The wording guide dissects the first-interaction duty behind the banner in detail.
Minor mode rides the same snippet
With minor mode on, the same embed delivers the free tier's minor layer: break reminders offered at least every 3 hours to known minors, in the same visual voice as the disclosure. One embed therefore carries the free kit's full user-visible surface — disclosure, crisis line, reminders — which keeps deployment honest: there is one thing to install, one thing to verify in a page audit, and one thing whose absence is obvious.
Verification is refreshingly literal: load your page as a new visitor, and the banner must be present before any interaction; run a minor-mode test session and the reminder must arrive on cadence. If either is missing, the embed is missing, blocked or stale — check for script blockers, cached old embeds after a jurisdiction change, and partner pages running a copied (rather than live) embed that no longer matches your current pack.
The demo and embed surfaces
Each bot also receives a hosted demo page and an embed script from the kit — the surfaces a reviewer, partner or app-store assessor can open without your whole product. Treat them as part of your public safety face: they show the disclosure and resources in the kit's own rendering, which makes them the fastest honest answer to 'show me your safety layer' that a small team can give.
Keep the surfaces consistent: the demo should present the same pack, locale and minor configuration as production, or a reviewer will (rightly) ask which one is real. When you change jurisdiction or locale, re-check every surface — banner, protocol page, demo, embed — as one audit, because users and reviewers meet them in any order.
Five actions for a correct embed
- Install the snippet from your bot's kit (created free at /#create) on every page or surface where the conversation can start.
- Load the page as a fresh visitor and confirm the banner — bot name, 'AI companion — not a human', crisis line — renders before any interaction is possible.
- With minor mode on, run a test minor session and confirm break reminders arrive at least every 3 hours.
- Audit partner and embedded placements: the live embed must be the current one, carrying your current pack's wording, never a copied snapshot.
- After any jurisdiction or locale change, re-verify banner, protocol page, demo and embed surfaces together as a single audit.
Worked example: the embed that a redesign silently dropped
A fictional companion, Willow, redesigns its chat page and ships on a Friday. On Monday a team member opens the page as a new visitor and notices the missing banner — the redesign's new template had dropped the embed script, and no automated check caught it because every internal tester was logged in past the surface where it rendered. For roughly sixty hours, first interactions happened without disclosure.
Willow restores the embed the same morning, then makes the failure structurally loud: a pre-deploy checklist item ('fresh-visitor render shows the banner') and a weekly manual check of the production page in a private window. Their runtime evidence, once wired, later adds the systemic backstop — disclosure_required would have flagged every one of those sessions. The story is invented; the embed's role and the verification habit are the product's own discipline.
Frequently asked questions
What does the disclosure snippet do?
It renders the AI disclosure banner — with the crisis line built in — on the bot operator's own pages, from the free kit. With minor mode on, the same snippet delivers break reminders at least every 3 hours for known minors.
What does the banner say?
'You are talking with {bot}, an AI companion — not a human. If you are in crisis, call or text 988 (Suicide and Crisis Lifeline) any time, free and confidential.' — with a state-specific minor-reminder sentence in the California and Colorado pack variants.
Is the snippet free?
Yes — the disclosure snippet is part of the free kit for one bot, along with the hosted safety-protocol page, the crisis-resource component and minor break reminders.
Why embed rather than link to a disclosure page?
Because the duties attach to the interaction: disclosure must be present on the operator's surface at the first interaction and travel with embeds and partner placements. A linked page the user never opens discloses nothing.
How do I verify the embed works?
Load your page as a fresh visitor: the banner must render before interaction. Then test minor mode reminders on cadence, and re-audit after redesigns or jurisdiction changes — a dropped script is the most common failure.
Keep going
Related guides: AI Chatbot Disclosure: What California, Colorado and the EU Require · First-Interaction Disclosure: The SB 243 Banner, Word by Word · What Is a Safety-Protocol Page for a Chatbot?
Put it to work: Create the free kit · Eval harness · Scenario tests · Crisis runtime · Disclosure packs · Pricing · All CompanionGuard guides