Trust Center
How we work, what we refuse, and what we cannot promise
A protection service asks for an unusual amount of trust. This page is the account of what we do with it — written to be read before you buy, not after something goes wrong.
Evidence
Evidence discipline
Every finding carries exactly one epistemic state, and promotion between states requires a recorded reason. A signal is not called an incident. A recommendation is not called a remediation. An observation without captured provenance is not called evidence. This constraint is enforced in the report structure itself, not left to the judgement of whoever is writing it.
Sources
Source provenance
Each finding records where it came from, when it was observed and how it was retrieved. Where a snapshot is lawful and technically possible it is captured at the moment of observation, because public content changes and disappears. If provenance could not be captured, the finding says so.
Authorisation
Customer authorisation
Your declared asset registry is your authorisation scope. We observe those assets and their public imitations, and nothing else. If something outside the scope appears relevant, we tell you it exists and ask whether to extend — extending is your decision and is recorded.
Data
Data minimisation
Onboarding collects only what is needed to scope and deliver your assessment: contact and billing details, the assets you want observed, your declared concern, your monitoring preferences and your authorisation. We do not ask for passwords, do not accept credentials, and do not require access to your systems.
Retention
Retention and deletion
Onboarding data and findings are retained while your engagement is active and for twelve months afterwards, so that a later report can reference an earlier baseline. You can request deletion at any time; we then remove your registry, onboarding data and findings, retaining only the minimal transaction record required for tax and accounting.
Access
Access control
Customer data is accessible to the operators delivering your engagement. Your workspace is reachable through your order reference together with the email used at checkout — that pairing is a convenience mechanism appropriate to the sensitivity of the data held, and the workspace deliberately holds no credentials, payment details or personal data beyond what you provided.
Accuracy
False positives and false negatives
Both occur. A look-alike domain may be entirely unrelated to you; an impersonating profile may exist in a source we did not observe. We manage this by reporting confidence honestly, by keeping unestablished findings visible as hypotheses rather than discarding them, and by stating what was not examined. We do not report a precision figure we cannot substantiate.
Automation
Limits of automated analysis
Observation and correlation are automated; the determination that something is a verified incident is not. An operator reviews every finding before it is described as verified. Automated similarity scoring is a measurement that directs attention — it is never presented as a conclusion.
Independence
Provider independence
We have no commercial relationship with the platforms, registrars or marketplaces where findings occur, and no incentive to over-report. We do not receive payment for volume of findings, and a report with no findings costs the same as one with twelve.
Audit
Auditability
Every report carries audit metadata: the observation period, the dimensions in and out of scope, the source families examined, and the operator review status. A report you received six months ago can be re-examined against the same metadata.
The assistant
The Guardian assistant
The assistant on this site is an automated guide. It identifies itself as automated, it operates from a fixed set of scripted paths, and it does not generate claims about your specific exposure. It cannot tell you that a scan has been run, cannot report findings, and cannot accept credentials. It exists to route you to the right offer and to help you through onboarding — nothing it says is a detection result.
Boundaries
Prohibited by default
Refused regardless of what a customer asks for or offers to pay.
- Unauthorised access to any account or system
- Intrusive network or vulnerability scanning
- Collection of credentials of any kind
- Biometric identification, or face-search claims without an explicit lawful basis
- Covert monitoring of private or access-controlled sources
- Autonomous legal notices
- Autonomous takedown requests absent a specific written mandate
- Contacting, pressuring or harassing an identified individual
- Fabricated urgency or invented findings
- Guaranteeing removal, or guaranteeing prevention
- Unsupported numerical, testimonial or customer claims
The signature
“Detect. Prevent. Protect.” is a purpose, not a guarantee
It describes what the service is organised to do. It is not a promise that every threat will be detected, that harm will be prevented, or that any specific asset will be protected. No observation service working from public sources can honestly promise those things, and we would rather say so on this page than in a disclaimer nobody reads.