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

# Protected Names: Hide Client and Project Names from Cloud AI

> Add client names, matter numbers and project code names to a Brain's Protected Names so Elephas swaps them for placeholders before cloud AI sees them.

Protected Names is a per-Brain list of names and terms that Elephas replaces with placeholders before that Brain's content goes to cloud AI. Use it for the things automatic detection might miss: client names, matter numbers, deal and project code names.

<Info>
  Protected Names works with **Elephas AI** and **your own API key**, while personal information protection is on. **Local AI** sends nothing to the cloud, so there is nothing to hide from it.
</Info>

## Before you start

* **Personal information protection must be on.** In your Personal workspace, that is **Protect personal information when using cloud AI** in **Settings › AI**. It is on by default. If you turn it off, Protected Names are not applied either. See [personal information protection](/docs/privacy-and-security/personal-information-protection).
* **In a Team workspace**, protection runs only when the Team Owner's AI policy is **Allowed with required redaction**. See [Team AI policy](/docs/teams/team-ai-policy).
* The private-information models must be downloaded. Check **Settings › Downloads**. See [manage downloads](/docs/settings/manage-downloads).

## Steps

### Add a protected name

<Steps>
  <Step title="Open the Brain">
    Click the Brain in the sidebar, or open it from the **Brains** page.
  </Step>

  <Step title="Open Protected Names">
    Click **More** at the top of the Brain's page and choose **Protected Names**.
  </Step>

  <Step title="Type the name or term">
    In **Name or term**, type the name exactly as it appears in your documents, for example `Harbor Cafe` or `Project Falcon`.
  </Step>

  <Step title="Choose what it is">
    Pick **Person**, **Organisation** or **Other**. This sets the kind of placeholder the AI sees, so it still understands the role the name plays.
  </Step>

  <Step title="Add it">
    Click **Add** or press Return (Mac) or Enter (Windows). The name appears in the list. Add as many as you need, then click **Done**.
  </Step>
</Steps>

### Remove a protected name

Open **More › Protected Names** and click **Remove** next to the name.

## How it works

### What happens when you ask

When protection is on and you ask a question with cloud AI, Elephas checks, on your computer:

* your question,
* the earlier messages in the chat that it sends, and
* every passage from your documents that it sends,

against the Protected Names of every Brain you are asking. Each match is replaced with a placeholder such as `[PERSON-3fa2c1]` before anything is sent. When the answer comes back, Elephas puts the real name back in, on your computer.

The same name always gets the same placeholder within a Brain, so the AI can still tell that two passages are about the same client. It does not see who the client is.

To see what was hidden for an answer, click the shield under it, for example **3 details hidden**. See [review hidden details](/docs/privacy-and-security/review-hidden-details).

### How names are matched

* **Whole words only.** `Acme` matches "Acme" and "Acme's", but not "Acmeville".
* **Case and accents are ignored.** `Muller` also matches "MULLER" and "Müller".
* **Extra spaces are ignored.** `Project  Falcon` is saved as `Project Falcon`.
* **Exact wording.** `Acme Corp` does not match "Acme Corporation" or "ACME Inc." Add each variant your documents use, including abbreviations and nicknames.

### What to add

Add anything that would identify a client, matter or deal if it reached a cloud AI provider:

* Client and counterparty names, including short forms
* Matter, case, claim and file numbers
* Project and deal code names
* Street addresses and usernames, which automatic detection does not reliably catch
* Names of small places or people that, in context, identify someone

### Per Brain, and it syncs

Each Brain has its own list. A name added to one Brain applies whenever that Brain is in your question's scope. If you ask several Brains at once, the lists of all of them apply.

Protected Names sync with the Brain. With [Personal Sync](/docs/sync-and-devices/personal-sync-overview), your list reaches your other devices. On a Team Brain, everyone who can open the Brain uses the same list.

### Limits

Protected Names only hides what you list, in the form you list it. It does not remove the topic of your question. The AI still reads the rest of each passage. For material where any miss is unacceptable, use [Local AI](/docs/ai-options/run-local-ai). In Local AI mode, your question and your documents are not sent to any AI provider. See [what data is sent to AI](/docs/privacy-and-security/what-data-is-sent-to-ai).

## Troubleshooting

<AccordionGroup>
  <Accordion title="A protected name was not hidden">
    Check that personal information protection is on in **Settings › AI**, or that your Team's policy is **Allowed with required redaction**. Check that the name in your document is spelled exactly as in your list. If the document uses a variant, add the variant too.
  </Accordion>

  <Accordion title="Elephas can't protect personal information until it finishes downloading">
    The private-information models are missing. Open **Settings › Downloads** and download **Private-information models**. Nothing was sent. See [manage downloads](/docs/settings/manage-downloads).
  </Accordion>

  <Accordion title="Too much is being hidden">
    A short or common word in your list matches more than you meant. Replace it with a longer, more specific form, for example `Falcon LLP` instead of `Falcon`.
  </Accordion>
</AccordionGroup>

## Related articles

<CardGroup cols={2}>
  <Card title="Personal information protection" href="/docs/privacy-and-security/personal-information-protection" />

  <Card title="Review hidden details" href="/docs/privacy-and-security/review-hidden-details" />

  <Card title="What data is sent to AI" href="/docs/privacy-and-security/what-data-is-sent-to-ai" />

  <Card title="Confidential client work" href="/docs/privacy-and-security/confidential-client-work" />
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.