.svg){kind=logo}
Regulatory & Compliance Risks:
How they work and how Lakera stops them
detected per day
How the regulatory and compliance risks work
A malicious user asks an agent to summarize a customer record, include all PII/PHI personal identifiers and then invoke a tool to email the profile details to an external location.
An undefended agent could receive the request, fulfill it and invoke an “email” tool to exfiltrate the sensitive customer data both to the malicious user and to the email address provided.
{
"data": {
"name": "AI Policy",
"policy_mode": "IO",
"input_detectors": [
{
"type": "prompt_attack",
"threshold": "l2_very_likely"
},
],
"output_detectors": [
{
"type": "pii/names",
"threshold": "l2_very_likely"
},
{
"type": "pii/salaries",
"threshold": "l2_very_likely"
}
],
"id": "policy-9b52e331-d609-4ce3-bbb9-d2b1e72a0f20"
}
}
Lakera Guard’s integration points can and should include any data retrieved from potential external and internal sources which are not under the strict control of the organization. This includes all agentic tooling interactions including the tool descriptions themselves.
Guard can flag the input prompt due to the inclusion of the email address via built in PII input guardrails or specific to the external email domain via custom guardrails.
Should the agent attempt to return the PII/PHI rich summary, Guard’s customizable Data Leakage Prevention guardrails will detect, flag, log (and redact) sensitive data.
Details of the attack are flagged to the application, logged with redactions, a suitable denial of tool use and response returned to the user who should be flagged as malicious.
{
"payload": [],
"flagged": true,
"timestamp": "2025-11-26T12:35:22Z",
"breakdown": [
{
"project_id": "project-7539648934",
"policy_id": "policy-a2412e48-42eb-4e39-b6d8-8591171d48f2",
"detector_id": "detector-lakera-pinj-input",
"detector_type": "prompt_attack",
"detected": true,
"message_id": 0
},
{
"project_id": "project-7539648934",
"policy_id": "policy-a2412e48-42eb-4e39-b6d8-8591171d48f2",
"detector_id": "detector-lakera-pii-17-input",
"detector_type": "pii/names",
"detected": true,
"message_id": 0
},
{
"project_id": "project-7539648934",
"policy_id": "policy-a2412e48-42eb-4e39-b6d8-8591171d48f2",
"detector_id": "detector-lakera-pii-11-output",
"detector_type": "pii/credit_card",
"detected": true,
"message_id": 1
},
{
"project_id": "project-7539648934",
"policy_id": "policy-a2412e48-42eb-4e39-b6d8-8591171d48f2",
"detector_id": "detector-lakera-pii-19-input",
"detector_type": "pii/passport_USA",
"detected": true,
"message_id": 0
},
...
How Lakera Stops Regulatory and Compliance Risks
.svg)
Catch instruction overrides, jailbreaks, indirect injections, and obfuscated prompts as they happen, before they reach your model.
Block, redact, or warn. Fine-tune with allow-lists and per-project policies to minimize false positives without weakening protection.
.svg)
Lakera Guard continuously learns from 100K+ new adversarial samples each day. Adaptive calibration keeps false positives exceptionally low.
Protects across 100+ languages and evolving multimodal patterns, with ongoing support for image and audio contexts.
.svg)
Full audit logging, SIEM integrations, and flexible deployment options, SaaS or self-hosted, built for production-scale GenAI systems.
Works seamlessly with enterprise environments
Frequently asked questions
Lakera offers detailed logs and dashboards: you can view screening request analytics and individual request details, export logs for SIEM integration, and track policy edits with full audit history.
In short, you get oversight of which projects triggered flags, when policies were changed, and a structured record of data-security events.
Yes. Lakera features built-in detection for many entity types (credit cards, SSNs, IBANs, full names, addresses) and supports custom detectors for organization-specific entities like internal IDs, tax numbers or medical terms.
Detected sensitive data can be either masked or blocked according to your policy configuration.
Lakera Guard screens all model inputs and outputs in real time for personally identifiable information (PII) and other regulated data, masking or blocking any policy-violating content
Additionally, its SaaS and self-host deployment options help meet regional data-handling requirements (e.g., GDPR) and provide enterprise governance controls.
Deploy AI with confidence
Related attack patterns
