|
| 1 | +--- |
| 2 | +pcx_content_type: how-to |
| 3 | +title: Set up Data Loss Prevention (DLP) |
| 4 | +sidebar: |
| 5 | + order: 2 |
| 6 | +--- |
| 7 | + |
| 8 | +Add Data Loss Prevention (DLP) to any AI Gateway to start scanning AI prompts and responses for sensitive data. |
| 9 | + |
| 10 | +## Prerequisites |
| 11 | + |
| 12 | +- An existing [AI Gateway](/ai-gateway/get-started/) |
| 13 | + |
| 14 | +## Enable DLP for AI Gateway |
| 15 | + |
| 16 | +1. Log into the [Cloudflare dashboard](https://dash.cloudflare.com/) and select your account. |
| 17 | +2. Go to **AI** > **AI Gateway**. |
| 18 | +3. Select a gateway where you want to enable DLP. |
| 19 | +4. Go to the **Firewall** tab. |
| 20 | +5. Toggle **Data Loss Prevention (DLP)** to **On**. |
| 21 | + |
| 22 | +## Add DLP policies |
| 23 | + |
| 24 | +After enabling DLP, you can create policies to define how sensitive data should be handled: |
| 25 | + |
| 26 | +1. Under the DLP section, click **Add Policy**. |
| 27 | +2. Configure the following fields for each policy: |
| 28 | + - **Policy ID**: Enter a unique name for this policy (e.g., "Block-PII-Requests") |
| 29 | + - **DLP Profiles**: Select the DLP profiles to check against. AI requests/responses will be checked against each of the selected profiles. Available profiles include: |
| 30 | + - **Financial Information** - Credit cards, bank accounts, routing numbers |
| 31 | + - **Personal Identifiable Information (PII)** - Names, addresses, phone numbers |
| 32 | + - **Government Identifiers** - SSNs, passport numbers, driver's licenses |
| 33 | + - **Healthcare Information** - Medical record numbers, patient data |
| 34 | + - **Custom Profiles** - Organization-specific data patterns |
| 35 | + |
| 36 | + :::note |
| 37 | + DLP profiles can be created and managed in the [Zero Trust DLP dashboard](/cloudflare-one/policies/data-loss-prevention/dlp-profiles/). |
| 38 | + ::: |
| 39 | + |
| 40 | + - **Action**: Choose the action to take when any of the selected profiles match: |
| 41 | + - **Flag** - Record the detection for audit purposes without blocking |
| 42 | + - **Block** - Prevent the request/response from proceeding |
| 43 | + |
| 44 | + - **Check**: Select what to scan: |
| 45 | + - **Request** - Scan user prompts sent to AI providers |
| 46 | + - **Response** - Scan AI model responses before returning to users |
| 47 | + - **Both** - Scan both requests and responses |
| 48 | + |
| 49 | +3. Click **Save** to save your policy configuration. |
| 50 | + |
| 51 | +## Manage DLP policies |
| 52 | + |
| 53 | +You can create multiple DLP policies with different configurations: |
| 54 | + |
| 55 | +- **Add multiple policies**: Click **Add Policy** to create additional policies with different profile combinations or actions |
| 56 | +- **Enable/disable policies**: Use the toggle next to each policy to individually enable or disable them without deleting the configuration |
| 57 | +- **Edit policies**: Click on any existing policy to modify its settings |
| 58 | +- **Save changes**: Always click **Save** after making any changes to apply them |
| 59 | + |
| 60 | +## Test your configuration |
| 61 | + |
| 62 | +After configuring DLP settings: |
| 63 | + |
| 64 | +1. Make a test AI request through your gateway that contains sample sensitive data. |
| 65 | +2. Check the **AI Gateway Logs** to verify DLP scanning is working. |
| 66 | +3. Review the detection results and adjust profiles or actions as needed. |
| 67 | + |
| 68 | +## Monitor DLP events |
| 69 | + |
| 70 | +### Viewing DLP logs in AI Gateway |
| 71 | + |
| 72 | +DLP events are integrated into your AI Gateway logs: |
| 73 | + |
| 74 | +1. Go to **AI** > **AI Gateway** > your gateway > **Logs**. |
| 75 | +2. Click on any log entry to view detailed information. For requests where DLP policies were triggered, additional details are included: |
| 76 | + - **DLP Action Taken**: Shows whether the action was "Flag" or "Block" |
| 77 | + - **DLP Policies Matched**: Detailed information about each policy that matched, including: |
| 78 | + - Which DLP profiles triggered within each policy |
| 79 | + - Whether the match occurred in the request or response |
| 80 | + - Specific entries that matched within each DLP profile |
| 81 | + |
| 82 | +### Filter DLP events |
| 83 | + |
| 84 | +To view only DLP-related requests: |
| 85 | + |
| 86 | +1. On the **Logs** tab, click **Add Filter**. |
| 87 | +2. Select **DLP Action** from the filter options. |
| 88 | +3. Choose to filter by: |
| 89 | + - **FLAG** - Show only requests where sensitive data was flagged |
| 90 | + - **BLOCK** - Show only requests that were blocked due to DLP policies |
| 91 | + |
| 92 | + |
| 93 | +## Error handling |
| 94 | + |
| 95 | +When DLP policies are triggered, your application will receive additional information through response headers and error codes. |
| 96 | + |
| 97 | +### DLP response header |
| 98 | + |
| 99 | +When a request matches DLP policies (whether flagged or blocked), an additional `cf-aig-dlp` header is returned containing detailed information about the match: |
| 100 | + |
| 101 | +#### Header schema |
| 102 | + |
| 103 | +```json |
| 104 | +{ |
| 105 | + "findings": [ |
| 106 | + { |
| 107 | + "profile": { |
| 108 | + "context": {}, |
| 109 | + "entry_ids": ["string"], |
| 110 | + "profile_id": "string" |
| 111 | + }, |
| 112 | + "policy_ids": ["string"], |
| 113 | + "check": "REQUEST" | "RESPONSE" |
| 114 | + } |
| 115 | + ], |
| 116 | + "action": "BLOCK" | "FLAG" |
| 117 | +} |
| 118 | +``` |
| 119 | + |
| 120 | +#### Example header value |
| 121 | + |
| 122 | +```json |
| 123 | +{ |
| 124 | + "findings": [ |
| 125 | + { |
| 126 | + "profile": { |
| 127 | + "context": {}, |
| 128 | + "entry_ids": ["a1b2c3d4-e5f6-7890-abcd-ef1234567890", "f7e8d9c0-b1a2-3456-789a-bcdef0123456"], |
| 129 | + "profile_id": "12345678-90ab-cdef-1234-567890abcdef" |
| 130 | + }, |
| 131 | + "policy_ids": ["block_financial_data"], |
| 132 | + "check": "REQUEST" |
| 133 | + } |
| 134 | + ], |
| 135 | + "action": "BLOCK" |
| 136 | +} |
| 137 | +``` |
| 138 | + |
| 139 | +Use this header to programmatically detect which DLP profiles and entries were matched, which policies triggered, and whether the match occurred in the request or response. |
| 140 | + |
| 141 | +### Error codes for blocked requests |
| 142 | + |
| 143 | +When DLP blocks a request, your application will receive structured error responses: |
| 144 | + |
| 145 | +- **Request blocked by DLP** |
| 146 | + - `"code": 2029` |
| 147 | + - `"message": "Request content blocked due to DLP policy violations"` |
| 148 | + |
| 149 | +- **Response blocked by DLP** |
| 150 | + - `"code": 2030` |
| 151 | + - `"message": "Response content blocked due to DLP policy violations"` |
| 152 | + |
| 153 | +Handle these errors in your application: |
| 154 | + |
| 155 | +```js |
| 156 | +try { |
| 157 | + const res = await env.AI.run('@cf/meta/llama-3.1-8b-instruct', { |
| 158 | + prompt: userInput |
| 159 | + }, { |
| 160 | + gateway: {id: 'your-gateway-id'} |
| 161 | + }) |
| 162 | + return Response.json(res) |
| 163 | +} catch (e) { |
| 164 | + if ((e as Error).message.includes('2029')) { |
| 165 | + return new Response('Request contains sensitive data and cannot be processed.') |
| 166 | + } |
| 167 | + if ((e as Error).message.includes('2030')) { |
| 168 | + return new Response('AI response was blocked due to sensitive content.') |
| 169 | + } |
| 170 | + return new Response('AI request failed') |
| 171 | +} |
| 172 | +``` |
| 173 | + |
| 174 | +## Best practices |
| 175 | + |
| 176 | +- **Start with flagging**: Begin with "Flag" actions to understand what data is being detected before implementing blocking |
| 177 | +- **Tune confidence levels**: Adjust detection sensitivity based on your false positive tolerance |
| 178 | +- **Use appropriate profiles**: Select DLP profiles that match your data protection requirements |
| 179 | +- **Monitor regularly**: Review DLP events to ensure policies are working as expected |
| 180 | +- **Test thoroughly**: Validate DLP behavior with sample sensitive data before production deployment |
| 181 | + |
| 182 | +## Troubleshooting |
| 183 | + |
| 184 | +### DLP not triggering |
| 185 | + |
| 186 | +- Verify DLP toggle is enabled for your gateway |
| 187 | +- Ensure selected DLP profiles are appropriate for your test data |
| 188 | +- Confirm confidence levels aren't set too high |
| 189 | + |
| 190 | +### Unexpected blocking |
| 191 | + |
| 192 | +- Review DLP logs to see which profiles triggered |
| 193 | +- Consider lowering confidence levels for problematic profiles |
| 194 | +- Test with different sample data to understand detection patterns |
| 195 | +- Adjust profile selections if needed |
| 196 | + |
| 197 | +For additional support with DLP configuration, refer to the [Cloudflare Data Loss Prevention documentation](/cloudflare-one/policies/data-loss-prevention/) or contact your Cloudflare support team. |
0 commit comments