Google analytics are off.
Shadow AI monitoring
Shadow AI is the use of AI applications, services, models or embedded features outside an organisation’s approved inventory and governance process.
Promptective brings recognised and suspected browser and desktop AI activity into one governance workflow, then shows which supported interactions are protected.
Direct answers
Shadow AI is the use of an AI application, service, model or embedded feature outside an organisation’s approved inventory and governance process. It can include public tools, AI features inside broader applications, code assistants, agents and direct model connections.
Combine an owner-reviewed AI register with signals from managed browsers and desktop devices. Promptective surfaces recognised AI applications and suspected connections so Security, Compliance and IT teams can confirm the owner, purpose and required policy.
Observed means Promptective detected evidence that an AI application or connection was used. Protected means Promptective checked a supported interaction before delivery, applied organisation policy and recorded the result.
Yes. Promptective can block supported AI traffic according to organisation policy while keeping prompt and response text out of routine records. Other discovered activity remains visible for review.
Record the AI application or feature, provider, owner, purpose, approval state, device, first and last seen dates, protection status and review history. Promptective keeps prompt, response and tool-result text out of routine inventory evidence.
Shadow AI risks
Assess the exact tool, account, data, permissions and business purpose before assigning risk.
Prompts, files and tool inputs can contain personal information, credentials, source code or commercial material that the organisation did not intend to share.
Retention, model-training, access, residency and deletion settings may not match organisation policy until the provider and account configuration are reviewed.
An unregistered use can lack an accountable owner, risk assessment, approved purpose, incident route, exception expiry or evidence for later review.
An AI feature, agent or connected tool with external permissions can read or change systems beyond its intended purpose. Risk depends on the exact identity, scope and action.
Observed versus protected
Promptective keeps newly discovered activity separate from supported interactions that were checked before delivery.
Promptective found recognised AI activity or a suspected connection that needs review.
Your team confirmed the application, provider, owner, business purpose and approval state.
Promptective checked the supported interaction, applied organisation policy and recorded whether content was delivered.
Inventory evidence
Keep enough evidence to explain the classification, protection and review without storing prompt or response text in routine inventory records.
Canonical application or feature, provider, account or workspace, accountable owner, users, business purpose and approval state.
The managed browser or device that produced the signal, its first and last seen dates and the review status.
Reviewed information classes, integrations, model or service origin, permissions and external actions.
Whether activity was discovered or protected, tied to the application, provider, device and active Promptective integration.
Policy and detector versions, decision, reason identifiers, transformations, delivery or execution result, and control-health evidence.
Risk reasons, treatment, exception and expiry, reviewer, incidents, last assessment and next review date, without routine prompt or response plaintext.
Promptective protection
Promptective applies the same organisation policy and decision record across supported browser, desktop, provider, code-assistant and custom-agent workflows.
Typed, pasted, dropped and uploaded content
Promptective checks supported browser AI interactions before content reaches the provider.
Explore browser protectionAI apps and code assistants
Promptective keeps most policy decisions on the device and protects supported desktop and command-line AI workflows.
Explore desktop protectionTool and action controls
Apply organisation policy before supported agents send content or execute external actions, with a receipt for each decision.
Explore agent securitySource transparency
Australian Government, National AI Centre
Office of the Australian Information Commissioner
Governance walkthrough
Tell us which browser, desktop and agent workflows your team uses. We will show how Promptective discovers activity, applies policy and records each decision.
