Security

Platform Security

• Encrypted connections and secure token handling.
• Role-based access control and audit logs.
• Rate limiting and abuse detection on critical endpoints.

Responsible Use

Only scan targets you own or have explicit written permission to test. Unauthorized testing is illegal and may result in account suspension.

Vulnerability Reporting

If you discover a security issue, contact Support with details and we will investigate.

Pentriarch Responsible Use and AI Safety Policy

Last updated: 7 February 2026

Pentriarch uses AI to help users plan security tests, interpret findings, and draft remediation guidance. This policy explains how AI is used safely and what we expect from users.

1) What AI does in Pentriarch

• Help translate goals into a safe test plan.
• Summarise scan outputs and evidence.
• Draft vulnerability explanations and remediation steps.
• Prioritise findings using context you provide.
• Generate structured outputs such as checklists and reports.

AI is designed to support human decisions, not replace them.

2) What we do not allow

• Scan or attack systems without authorization.
• Steal data, bypass access controls, or enable wrongdoing.
• Create malware, exploit kits, or instructions for harm.
• Conduct harassment, doxxing, or surveillance.
• Violate laws, contracts, or third party rights.

We may suspend accounts involved in prohibited activity.

3) Safety by design

• Scope controls (targets, profiles, and execution modes).
• Tool guardrails for high risk actions.
• Logging and audit trails.
• Isolation for scan execution where feasible.
• Rate limits and abuse detection.
• Customer configurable allowlists and policies where available.

4) How we handle AI outputs

• Validate outputs before acting.
• Treat remediation guidance as suggestions, not guarantees.
• Use peer review for high risk changes.
• Confirm legal authorization and scope before running tests.

Pentriarch may label AI generated content and maintain links to evidence where possible.

5) Model providers and data handling

• Send only what is needed to complete the request.
• Encourage users not to submit secrets unless necessary.
• Provide options to redact or minimise sensitive inputs where available.

If you need strict handling (for example no external model processing), use enterprise controls or self hosted execution options where available.

6) Customer responsibilities

• Ensure they have permission to test targets.
• Ensure scans comply with internal policy and law.
• Configure role access, retention, and guardrails.
• Review outputs before acting on them.

7) Reporting and enforcement

If you believe AI behavior is unsafe or a user is abusing the platform, contact: security@pentriarch.com and privacy@pentriarch.com.

• Limiting features or execution modes.
• Blocking targets or payload patterns.
• Suspending accounts.
• Preserving evidence where legally required.

8) Continuous improvement

• User feedback.
• Incident reviews.
• Security testing.
• Updates to tool allowlists and risk policies.