Usage policy
Rules for using Cyberdroid AI.
The rules in brief
Decisions about people rest on evidence a person has reviewed, and Cyberdroid output goes to a person before it leads to an action.
Scope
This policy applies to Cyberdroid AI Detection, available now to select organizations, and Cyberdroid AI Investigation, in early access. It applies to every licensed organization and every person who uses the products on its behalf, alongside the organization's agreement with ARRTECH.
Findings are evidence
A score ranks the queue, and a graph relationship does not prove intent. An analyst reviews and accepts each case.
No decision about a person may rest solely on a score, a graph relationship or a case verdict. That includes discipline, dismissal, legal action, a report to an authority, or removing access other than as a security containment step. A person reviews the evidence first.
Review before action
Cyberdroid output goes to a person for review before it leads to an action in the environment, whether directly or through other software. DLP policies and SOAR automations the organization writes are outside this clause.
Lawful monitoring
The licensed organization is responsible for having the right to collect and analyze its SIEM's telemetry. It is also responsible for informing the people concerned where law or agreement requires it, and consulting employee representatives where law requires it. The products may be used to monitor people only to secure systems the organization is responsible for.
Access controls stay intact
No one may bypass the controls governing requests and recipients or impersonate an authorized requester. Access may be granted only to people entitled to receive it. Anyone who finds a way around one is asked to report it under Responsible disclosure.
External fallback
AI Investigation uses local models by default. The online fallback may be turned on only for data the organization may send to its provider.
Early access
AI Investigation may be used only within the terms of the organization's design-partner evaluation.
Reporting and enforcement
Suspected misuse is reported through our contact page, and an incorrect or harmful output through AI incident reporting. A breach of this policy is handled under the organization's agreement with ARRTECH. Each version carries an effective date.
