02 Operator
Who is accountable and how to reach us.
Operator
BlueLabs3, trading as 1HBT. ACN 701 379 964.
admin@consensor.io · zfa@consensor.io · privacy@consensor.io
03 Processing
What is processed and kept.
Processing
A caller may send a public key and a text field. Those bytes are processed to decide whether the request may proceed. They are not sold. The sealed ledger keeps a bounded trailing window. Hosting is at the edge. Recipients of those bytes are likely in the United States.
Ledger window is bounded. Sessions are in-memory and die with the isolate. What each layer keeps, and for how long, is listed on the trust page.
TLS is terminated at the edge.
Privacy requests: privacy@consensor.io. Security reports from named clients: zfa@consensor.io.
04 Not claimed
No certification or invite.
What is not claimed
No SOC 2, ISO, defence or clinical status is claimed on this host. A provisional patent application (AU 2026905672) is pending; no patent has been granted. Public testing is closed. There is no public invite. People will still try. That is expected and not invited.
Refused, malformed, and unauthenticated probe traffic is recorded and treated as hostile.
05 Accountability
Automated decisions and governance.
Accountability and automated decisions
BlueLabs³ (1HBT) is accountable for this gate. The Director is the accountable person; there is no delegate. Governance follows the Australian Voluntary AI Safety Standard read with the Guidance for AI Adoption. The BL3 Policy Manual (an index and fourteen policies, adopted 21 September 2026) governs security, risk, change, testing, incidents, oversight, transparency, redress, records, privacy, vendors and continuity. No SOC 2 report or certification is claimed.
Refusal decisions are automated and are not made by a language model. Answers are written by a language model and checked before they are returned, and every answer says so. The checks catch what they are built to catch: they can miss harmful content or refuse harmless content, and a model answer can be wrong. Do not rely on an answer for medical, legal, financial or safety decisions.
06 Challenge
How to challenge a decision.
Challenge a decision
Anyone affected by this gate can challenge the outcome, or the gate's use of AI itself, and ask for it to be put right: a person whose message was refused or answered through it, a named client, or anyone who objects to AI being used. Email admin@consensor.io with the subject Challenge, the date and time, the product you were using, what happened and what you want done. If email does not work for you, say so in any message or ask the named client to raise it for you, and we will arrange another way. The Director reviews every challenge personally; no challenge is decided by the gate or by a model. We acknowledge within 2 business days and decide within 10, in plain language, with reasons. Outcomes can include a correction or retraction of an answer, an explanation of why the gate acted as it did, a rule change under change control (you are told when it ships), notice to the named client, or, where the gate caused harm, a remedy agreed with you. If you disagree with the decision, reply and the Director reviews it again within 10 business days. Crisis refusals stay in place for everyone; the safeguard is not switched off for one person. Your complaint is still reviewed by a human, and you receive the reasons, support services and any remedy that does not disable the safeguard. We do not store your message, so include your own copy if you want it considered. This channel is read on business days and is not a crisis service: if you need help now in Australia, call Lifeline 13 11 14, or 000 in an emergency. You may also take a privacy matter to the Office of the Australian Information Commissioner, an online safety matter to the eSafety Commissioner, a consumer matter to your state or territory fair trading body, or a complaint to the named client through whose product you reached this gate.