> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trypillow.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI disclosure banner

> Let respondents know they're talking to an AI, as required by the EU AI Act and several US states.

When the **AI disclosure banner** is on, every respondent sees a notice at the top of the chat telling them the interview is conducted by an AI researcher. Your agent is also forced to confirm its AI nature if asked, so the message in the UI and the agent's answers stay consistent.

<Info>
  Several regulations, including the **EU AI Act** (Article 50) and state-level laws in California, Utah, and Colorado, require you to inform people when they are interacting with an AI system. Enabling this banner is the easiest way to stay compliant.
</Info>

## What respondents see

With the banner on, the chat opens with a short notice above the first message:

> *This interview is AI-moderated, it remains anonymous so feel free to give your honest feedback.*

The notice sits at the top of the interview and scrolls with the messages. It replaces the standard anonymity line, so respondents still get the "stays anonymous" reassurance, now bundled with the AI disclosure.

## What changes behind the scenes

Turning the banner on has two effects:

1. The notice appears at the top of the chat.
2. The agent's **reveal policy** is locked to **Reveal AI Identity**. If a respondent asks "are you a real person?", the agent answers truthfully.

Locking the reveal policy keeps the UI and the interview aligned. You cannot show a disclosure banner and have the agent deny being an AI at the same time.

## When to enable it

<CardGroup cols={2}>
  <Card title="EU respondents" icon="flag">
    The EU AI Act requires disclosing AI interaction to end users. Turn the banner on for any study reaching EU audiences.
  </Card>

  <Card title="US respondents" icon="flag">
    California, Utah, Colorado and other states require similar disclosures. Turning the banner on satisfies these requirements.
  </Card>

  <Card title="Sensitive topics" icon="shield">
    For studies touching health, finance, or other regulated areas, disclosure builds trust and reduces legal risk.
  </Card>

  <Card title="Public-facing studies" icon="globe">
    Whenever you're not sure who will take part, enabling the banner is the safe default.
  </Card>
</CardGroup>

## Turn it on or off

<Steps>
  <Step title="Open the agent settings">
    From the study builder, open the **Agent** drawer.
  </Step>

  <Step title="Find the Reveal policy section">
    Scroll to **Reveal policy**. The **Show AI disclosure banner** toggle sits at the top of this section.
  </Step>

  <Step title="Toggle it on">
    The **Reveal policy** select locks to **Reveal AI Identity** (that's expected).
  </Step>

  <Step title="Save">
    Click **Save**. The next time a respondent opens the chat, the notice is there.
  </Step>
</Steps>

<Tip>
  New studies have the banner **on by default**. Studies created before this feature shipped keep it off, so turn it on manually if you need disclosure.
</Tip>

## Defaults

| State                             | Banner | Reveal policy                    |
| --------------------------------- | ------ | -------------------------------- |
| New study                         | On     | Reveal AI Identity (locked)      |
| Study created before this feature | Off    | Keeps its previous value         |
| Banner turned off                 | Off    | Editable; pick any reveal policy |

## Frequently asked questions

<AccordionGroup>
  <Accordion title="Can I change the wording of the banner?">
    Not today. The wording is standardized across studies to keep disclosure clear and legally defensible. The text is automatically translated to match your respondent's language.
  </Accordion>

  <Accordion title="Does the banner appear in test mode too?">
    Yes. Test and live interviews both show the banner when it's on, so what you test is what respondents see.
  </Accordion>

  <Accordion title="Why is the Reveal policy locked when the banner is on?">
    If the banner says the interview is conducted by AI but the agent denies being an AI when asked, you create a contradiction that can undermine compliance. Locking the reveal policy keeps both layers honest.
  </Accordion>
</AccordionGroup>

## What's next

<CardGroup cols={2}>
  <Card title="Test vs. live mode" icon="flask-conical" href="/guides/test-vs-live">
    Verify the banner renders correctly before launching.
  </Card>

  <Card title="Multi-language support" icon="languages" href="/guides/multi-language">
    See how the disclosure is translated for international respondents.
  </Card>
</CardGroup>