---
title: Asset Remediation
description: Allow, warn, or block an AI asset. What each action does on the endpoint, and how to apply one to one asset, many assets at once, or one asset on one endpoint.
---

Every AI asset starts **Ungoverned**: found, scored, but nobody has taken action on it yet. It waits as an asset incident on **Incidents › Open**. When an Admin or Owner applies Allow, Warn, or Block, the asset becomes **Governed** and the incident resolves. See [Incidents](/surface/incidents/overview).

```mermaid
flowchart LR
  N[We find an asset] --> O[Open incident]
  O -->|Allow| A[Allowed]
  O -->|Warn| W[Warned]
  O -->|Block| B[Blocked]
```

## The three actions

| Action | What happens on the endpoint | When to use it |
| --- | --- | --- |
| **Allow** | Nothing changes. The asset is recorded as approved | Assets you are fine with |
| **Warn** | The person sees a system notification that the asset is not approved, and sees it again each time the asset opens. The asset stays installed | Assets under review, or tolerated with a reminder |
| **Block** (Beta) | We back the asset up, then remove it. If it cannot be removed, we close it every time it opens | Assets that are not permitted |

An action is company-wide by default. You can also make an exception for one endpoint, for example a security researcher who needs a blocked asset. See [One asset on one endpoint](#one-asset-on-one-endpoint).

:::info
Block is in beta. We're still improving it, so try it on a small group of endpoints before you apply it to everyone.
:::

## How an action reaches endpoints

```mermaid
sequenceDiagram
  participant Admin
  participant Console
  participant Endpoint
  Admin->>Console: Block "Asset X"
  Console-->>Endpoint: Notified within seconds (if online)
  Endpoint->>Endpoint: Back up, then remove Asset X
  Endpoint-->>Console: Report: removed or failed
  Console-->>Admin: Audit Log entry
```

Endpoints that are online act within seconds. Endpoints that are offline act at their next scan. Every endpoint reports back, and failures show in the [Audit Log](/surface/activity/audit-log). See [On-device enforcement](/surface/policies/on-device-enforcement).

:::warning
Block removes software from people's endpoints. Read the confirmation window. With **Global** it says "all affected endpoints" and gives no number. For a new asset, try Warn on a small group first.
:::

## Before you begin

You have the **Owner** or **Admin** role. Auditors can see the assets but not take action.

## Where you take action

Two paths lead to the same controls:

- **Incidents › Open**, on the **Assets** tab — every asset still waiting for a decision, highest risk first.
- An asset's panel in **Inventory** — the **Take Action** button at the bottom.

An asset stays governed even if no endpoint has reported it recently; actions never quietly disappear.

See [Resolve incidents](/surface/incidents/resolve) for the step-by-step: take an action on one asset's incident, apply one action to many selected assets at once (up to 500 per batch, highest risk first, all-or-nothing), or scope an action **By Instance** to named endpoints. Selecting assets and then changing tabs, filters, search, or sort asks before the selection is discarded.

## One asset on one endpoint

To exempt one endpoint, use **By Instance** when you take the action, or start from the endpoint: on **Endpoints**, open the endpoint, find the asset on its **Assets** tab, and use **Allow**, **Warn**, or **Block** (Beta) in its panel — that applies to this endpoint only.

For a **Governed** asset with no exception on this endpoint, the panel shows **Inherits Default** under **Current Enforcement**: the endpoint follows the company-wide action. For an **Ungoverned** asset, the panel shows **Recommended Action** instead.

## Checking what was applied

Every action, single or batch, is in the [Audit Log](/surface/activity/audit-log) under **Asset Governance**, with who applied it, when, and the before and after state of each asset.

## Next steps

- [Incidents](/surface/incidents/overview) for the list every ungoverned asset waits in
- [Audit Log](/surface/activity/audit-log) to see what was applied and when
- [Endpoints](/surface/environment/endpoints) to check an endpoint after a Block
- [How risk is calculated](/surface/overview/risk-scoring) to choose with the score in mind
