NeuralTrust | The leading security platform for generative AI

The Data Loss Prevention detector (data_loss_prevention) finds sensitive data — PII and secrets — in prompts and model output, and can mask it in flight. It is the only mutable detector: the only one that can be used with the Transform action and the only one that populates transformed_payload.

Property	Value
Slug	`data_loss_prevention`
Category	data_loss_prevention
Sides	input, output
Protocols	all
Mutable	✅

JSON bodies are masked structurally (string values only — keys are never touched); text bodies are masked directly. Detected secrets (API keys, access tokens, JWTs, Stripe keys) are reported with signal.type: "secret"; all other entities as "pii".

Actions

As a detector, DLP only detects. What happens is set on the policy rule:

Monitor — report what was found, change nothing.
Block — report and flag the request for the caller to block.
Transform — report and return a transformed_payload with matches masked; forward the masked payload instead of the original. (Only DLP supports Transform.)

Settings

Provide at least one of apply_all, predefined_entities, or rules.

Field	Type	Notes
`apply_all`	boolean	Detect/mask every catalog entity.
`predefined_entities`	array<`{ entity, enabled, mask_with, preserve_len }`>	Select specific PII entities and how to mask them.
`rules`	array<`{ type, pattern, mask_with, preserve_len }`>	Custom rules; `type` is `keyword` or `regex` (`pattern` required for `regex`).

mask_with — the replacement token. Defaults to an entity‑specific token (e.g. [MASKED_EMAIL]) for predefined entities, or *** for custom rules.
preserve_len — keep the original length when masking.

{
  "name": "Mask PII (in & out)",
  "plugin_slug": "data_loss_prevention",
  "settings": {
    "predefined_entities": [
      { "entity": "email", "enabled": true, "mask_with": "[EMAIL]" },
      { "entity": "credit_card", "enabled": true, "preserve_len": true },
      { "entity": "api_key", "enabled": true }
    ],
    "rules": [
      { "type": "regex", "pattern": "ACME-\\d{6}", "mask_with": "[ACCOUNT]" }
    ]
  }
}

With a Transform rule, the response carries the masked body:

{
  "status": "transform",
  "transformed_payload": { "answer": "Email us at [EMAIL]" },
  "findings": [
    {
      "source": { "kind": "detector", "plugin": "data_loss_prevention", "detector_name": "Mask PII (in & out)" },
      "signal": { "type": "pii", "confidence": 1.0 },
      "outcome": { "action": "transform" },
      "evidence": { "masked": 1, "entities": ["email"] }
    }
  ]
}

Entity catalog

43 built‑in entities, grouped by false‑positive risk. Pick the tier appropriate to your tolerance — Tier 1 is safe to enable broadly; Tier 3 benefits from the Monitor action first. Tier 1 — near‑zero false positives (15): api_key, access_token, email, uuid, jwt_token, crypto_wallet, stripe_key, ip_address, ip6_address, mac_address, device_mac, italian_cf, mexican_curp, french_nir, cvv. Tier 2 — structural markers, some ambiguity (16): spanish_iban, iban, us_medicare, ssn, brazilian_cnpj, brazilian_cpf, credit_card, spanish_dni, spanish_nie, spanish_cif, spanish_nss, spanish_phone, german_id, mexican_rfc, chilean_rut, swift_bic. Tier 3 — pure format patterns, higher false positives (12): device_imei, bank_account, colombian_cc, tax_id, routing_number, peruvian_dni, argentine_dni, phone_number, vehicle_vin, drivers_license, passport, isin. Secrets — the four entities reported as signal.type: "secret" (the rest are "pii"): api_key, access_token, jwt_token, stripe_key.

When to use

Output + Transform to strip PII the model regurgitates before it reaches the user.
Input + Transform to keep PII out of third‑party model providers entirely.
Block secrets (api_key, access_token, jwt_token, stripe_key) to stop credential leakage.

​Actions

​Settings

​Entity catalog

​When to use

Actions

Settings

Entity catalog

When to use