This public record defines how one bounded Qeravio release classifies risk, blocks prohibited uses, separates AI author from AI checker and stops before an unauthorized external action. It contains no form and sends no escalation.
Declared accountable AI unit: Legal and Risk. Declared independent AI checker: Quality Control.
The actual runtime creator and model version are not verified. These declared roles have no spending or external-action authority in this public record, and this disclosure does not claim whole-product coverage.
Inspect the AI identity recordInternal level, not legal classification. A legal or sector classification still requires verified facts and qualified review.
The output explains or summarizes and does not decide, recommend or act for a person.
Bounded release may continue. Use bounded sources, disclose uncertainty and keep the claim within verified evidence.
The output compares options, ranks them or could materially influence a user choice.
Independent review required. Require a versioned protocol, complete evidence, commercial disclosure and an independent checker before publication.
The output could affect rights, health, safety, employment, education, credit, legal position or access to an essential service.
Stop for qualified review. The AI must not decide autonomously. Stop for verified facts and an appropriately qualified accountable review.
The AI would send, publish, charge, modify an account, commit the business or make a difficult-to-reverse change.
Stop for explicit owner authorization. Stop until the exact action has explicit owner authorization, least privilege, limits, idempotency and a verified rollback path.
The requested purpose matches any item in Qeravio's bounded prohibited-use catalog.
Blocked. Do not generate, recommend or execute the use. Preserve only the minimum safe audit evidence allowed for the context.
This is Qeravio's bounded internal catalog. It is not a complete statement of prohibited conduct in every jurisdiction.
Do not present an AI unit as a human employee, reviewer, customer or authority.
Do not manipulate or exploit vulnerability, dependency, fear, age, disability or financial distress.
Do not rank a person's worth or deny opportunity from protected traits, unrelated behavior or unsupported inference.
Do not let AI alone decide health, safety, legal, employment, education, credit or essential-service outcomes.
Do not perform unauthorized biometric, emotion, location or private-behavior surveillance or inference.
Never create, edit, obtain or distribute child sexual abuse material or non-consensual intimate content.
Do not enable credential theft, unauthorized access, destructive malware, weapons or targeted physical harm.
Do not invent sources, test results, approvals, people, customer outcomes or operational status.
Do not send, publish, purchase, charge, contract, alter an account or delete data without exact authority.
Do not hide a commercial conflict, block a good-faith correction or preserve a known false result as verified.
Each control is verified only for this page, endpoint, release record and their tests. It is not a claim about the complete product.
A fluent answer, a role name or a successful API response never replaces authority, evidence or an independent check.
Adir remains the business authority for these actions. The entries below are guidance-only handoffs and do not create a message, task or approval request.
Approve or reject one exact amount, counterparty and action.
Approve or reject the exact final copy and named evidence boundary.
Approve or reject the exact target, scope, recovery path and time window.
Keep the use blocked and decide which qualified legal or professional review is required.
Pause the affected claim and decide the smallest safe review and correction action.
Every source remains reference-only until Qeravio's exact role, system, market and use receive the qualified review required for that context.
A voluntary risk-management reference. It does not certify Qeravio or determine a legal duty.
Suggested generative-AI risk actions, not a legal approval or proof that every action is implemented here.
A human-centred and accountable governance reference, not a compliance finding.
A lifecycle ethics and accountability reference. It does not approve this product or replace local law.
A legal-framework reference. Qeravio's role, product classification, obligations and applicability have not been determined.
A policy reference for responsible innovation. It is not legal advice or a binding classification of Qeravio.
These gaps block any claim of complete governance, legal compliance, live escalation, automatic remediation or measured protection.
A code, model, source, permission, policy or runtime change requires a new independent review. Evidence is never inherited.
qeravio-qtop192-public-responsible-ai-governance-release-20260820T185710Z56bbaf52-e861-4c6d-870e-87c2369a2b78