Trust Centre

Security and privacy, built into where PromptLeash runs.

PromptLeash runs inside your environment, under your security standards. Updates ship one way, from us to you, and nothing comes back. This is where your security, privacy and procurement teams can review how it works.

Book a demoRequest the security pack

At a glance

The deployment model in four facts.

  • Deployed in your private environment

    PromptLeash is deployed inside your environment, works under your security standards and has no external dependencies.

  • No data leaves

    No telemetry, no usage data and no interaction content leaves your environment or reaches PromptLeash.

  • Updates ship one way

    Updates travel from PromptLeash to you, and you control deployment and updates.

  • Your identity and access

    Single sign-on and role-based access work within the configuration agreed for your deployment.

Architecture

How data moves, and where it stops.

PromptLeash architecture and data flowApproved AI sources feed activity into your environment. PromptLeash classifies and aggregates it there, under your identity, access and logging, and produces views for executives, managers and individuals. Updates ship one way from PromptLeash, and no data leaves your environment.Your environmentYour security standardsApproved AI sourcesProviders, chat assistants,applications and agentsAgreed scopePromptLeash analysisClassificationAggregationFindingsRole-based viewsExecutiveManagerIndividualYour identity, access and loggingPromptLeashBuilds and ships updatesUpdates, one wayNothing comes back

Step by step

  1. Approved AI sources, such as providers, chat assistants, applications and agents, connect inside your environment. Only activity within the agreed scope is read.
  2. PromptLeash classification and aggregation run in your environment, separately from live AI requests.
  3. Everything runs under your identity, access and logging, and your security standards.
  4. The results become role-based views for executives, managers and individuals, each showing what that audience should see.
  5. PromptLeash ships updates one way, into your environment. Nothing comes back: no telemetry, no usage data and no interaction content.

Data location

What stays, and what reaches us.

What stays in your environment and what reaches PromptLeash
Stays in your environment
Prompts, responses and agent activity from your approved AI sources, within the agreed scope
Usage and activity records
Classification results and aggregated evidence
Your role and workflow context
Executive, manager and individual views
Identity, access and logging, in your own systems

Data handling

How PromptLeash handles your data.

What does PromptLeash read?
Approved AI activity from the sources you put in scope, including interaction content such as prompts, responses and agent activity, within the agreed scope. Nothing is analysed until it is in the agreed scope.
Where is it processed?
In your environment. Classification and aggregation run there, alongside your AI stack rather than in the path of a live request.
Who governs retention and deletion?
You do. Retention and deletion follow your own standards.
Who controls identity and access?
You do, under your own standards. Single sign-on and role-based access work within the configuration agreed for your deployment.
Who sees the evidence?
Each audience sees its own view. Individuals see their own results, and manager and executive views are aggregated.
Which AI sources can it connect to?
Any AI source you approve, including providers, chat assistants, applications and agents. The sources connected in your deployment are agreed during scoping.
Does anything reach PromptLeash?
No. Nothing comes back to PromptLeash: no telemetry, no usage data and no interaction content.
How employee privacy works

Assurance

Where our assurance work stands.

Status at 5 October 2026
AssuranceStatus
SOC 2An independent auditor's report on PromptLeash's security controls.In progress,
targeted for the end of 2026
Independent penetration testSecurity testing of PromptLeash by an independent tester.Planned,
targeted for the end of 2026

This board lists assurance work under way. We update it as each step completes, and no badge appears until the work is complete.

Found a security issue? Read our vulnerability disclosure policy.

Responsible AI

Responsible AI, by design.

PromptLeash is a finalist for Best Application of Responsible AI at the Australian AI Awards 2026. These are the principles behind that work.

Read the announcement

  1. 01

    Data stays under your control.

    PromptLeash runs inside your environment and analyses usage data there, rather than sending it to us. Analysis runs separately from live AI requests, so it never sits between your people and the tools they rely on.

  2. 02

    Transparency leaders can act on.

    PromptLeash shows whether AI use follows your policies as part of adoption, so risk and value are visible in the same place.

  3. 03

    Measurement in service of people.

    Measuring AI use should help people use it well: showing where teams need support, where training should be targeted and where people are already doing excellent work that others can learn from. Access to insights is controlled by role.

In the Trust Centre

Read the detail.

Take PromptLeash through your review.

Book a demo to see PromptLeash with your team, or request the security pack for your security review.

Book a demoRequest the security pack