Trust
Trust is built from rules you can check
We test real systems, so the rules matter more than the promises. This page sets out how we authorize testing, keep it within limits, handle your data and choose AI models, and who is accountable for each engagement.
Safe by construction, not by promise.
These rules are enforced by the platform and followed by the team. They are why we can test production systems continuously.
-
Ownership first
Nothing beyond passive analysis runs until you have verified that you control the domain, with a DNS TXT record or a verification file.
-
Every action has a scope
Each automated action is tied to an approved scope and policy, and checked before it runs.
-
Exclusions always win
If a policy allows something your scope excludes, the exclusion wins.
-
Approval for elevated steps
Vulnerability probes need your explicit approval. Automated exploitation is not allowed in any mode.
-
An append-only record
Significant actions and decisions are recorded in an append-only timeline you can review.
-
Secrets stay masked
Secret values found during testing are masked in findings, reports and notifications. Full evidence stays in private, access-controlled storage.
Testing runs only in the mode you approve.
Four modes, from public sources only to operator-assisted testing that needs your explicit approval. No mode allows automated exploitation.
| Mode | Risk | What it can do | What it never does | Approval |
|---|---|---|---|---|
| Passive analysis | Minimal | Public sources only: DNS, WHOIS, certificate transparency logs, open-source intelligence, credential-exposure checks | Sends no requests to your systems | Not required |
| Safe validation | Low | Adds standard HTTP requests, TLS and header review, and TCP port checks | No vulnerability probes, no fuzzing, no authenticated testing | Not required |
| Focused investigation | Moderate | Adds endpoint probing, sensitive-path checks and technology fingerprinting | No fuzzing, no authenticated testing, no exploitation | Not required |
| Operator-assisted | Elevated | Adds vulnerability probes | No exploitation | Your explicit approval |
People who answer for the work
Every engagement is directed by our security director, who holds a postgraduate degree in information security and cryptography, a master’s degree in cyber defense and more than 20 years of experience. The rest of the team monitors, follows and works every engagement, and the director signs our reports. Our methodology is shaped by years of hands-on offensive research and responsible disclosure.
How we handle your data
-
Your findings are yours
Results, vulnerability data and engagement history belong to your organization.
-
No data mining
Platform telemetry is limited to operational metrics. Your targets and findings are never analyzed for commercial purposes.
-
Deletion on request
You can request permanent deletion of your scan data, findings and engagement history at any time.
-
Isolation between clients
Each organization’s data is isolated. Findings from one organization are not accessible to another.
-
Sensitive evidence kept private
Secret values are masked in findings, reports and notifications. Full evidence stays in private storage, retrievable only by authorized users.
-
Controlled exports
Report exports and API access are governed by role-based access controls.
Your data, your jurisdiction, your choice of model.
-
Data residency
We host your data where you, or your regulators, need it: for example in US or EU data centers.
-
Model choice
We are model-agnostic and choose the model per client need, with access to the latest frontier models. If you require it, we use our proprietary, closed model, which runs fully isolated from commercial AI providers.
-
Your findings are yours
Results and evidence belong to your organization. We do not analyze them for commercial purposes, and you can ask us to delete them at any time.
Subprocessors and data agreements
We do not sell your data or your findings. Some data may be shared with subprocessors needed to operate the service, under data processing agreements; if you require that no commercial AI provider processes your engagement data, we use our isolated proprietary model. For custom data processing agreements or specific jurisdictional commitments, write to enterprise@northboundintelligence.com. Privacy questions go to privacy@northboundintelligence.com, and our Privacy Policy explains how we process personal data.
Every report we issue can be verified
Every PDF report we issue is digitally signed with our document-signing certificate, which chains to the NorthBound Intelligence Root CA. The signature lets anyone confirm that nothing was changed after signing, and PDF readers flag any later edit. Each report also carries a timestamp from an independent timestamping authority, a document ID on every page and a NorthBound watermark, and its metadata contains only what we put there. To check a report, upload it or enter its document ID at app.northboundintelligence.com/verify. You can also add our root certificate to your PDF reader’s trusted certificates; its SHA-256 fingerprint is 1A:17:25:1F:AF:3A:71:28:5A:9D:07:80:CB:F4:39:E4:75:31:00:4E:5B:A5:74:12:30:68:D1:DF:0E:C8:2E:0A.
Found a vulnerability in our systems?
Tell us. Our vulnerability disclosure page explains how to report it, and our contact details are also published in security.txt.
Questions about trust and data
Where is our data stored?
Where you or your regulators require, for example in US or EU data centers.
Can we require that no commercial AI provider sees our data?
Yes. For clients who require it, we use our proprietary, closed model, which runs fully isolated from commercial AI providers.
Who can see our findings?
Authorized users of your organization and the NorthBound team working on your engagement. Access is role-based.
How do we have our data deleted?
Ask us at privacy@northboundintelligence.com. You can request permanent deletion at any time.
Find it before an attacker does.
Tell us what you need covered. We will reply from contact@northboundintelligence.com to arrange a scoping call.
Facts on this page verified 11 November 2026.