What EthosGuard is not
A tool that evaluates other people's AI should be clear about its own limits. Here they are, before anyone has to discover them.
Not a certification body
EthosGuard evaluates AI answers, suggests revisions, routes findings to people and keeps the record. It does not certify, issue a mark, keep a public register, or decide whether an organization meets a legal standard. Where a client wants an independent mark, we point them to a certification body and take no part in its decision.
Not professional advice
EthosGuard does not provide medical, legal or financial advice. A suggested revision is a draft for a qualified person to review. The named reviewer, not the tool, answers for what is sent.
Not proof of compliance
EthosGuard keeps evidence that supports governance work under frameworks such as the EU AI Act, NIST AI RMF and ISO/IEC 42001. It does not by itself establish that an organization complies with any of them.
Not a verdict without reasons
Every score and flag comes with written reasoning tied to a named dimension. A finding that cannot be explained in plain language counts against us. The Methodology page says how to raise it.
Not a filter on all your traffic
In a pilot, EthosGuard reads the answers of one agreed workflow in monitor mode, alongside production. It does not block or replace answers unless that is agreed later, on the evidence.
Not finished
Some capabilities run today, some only inside a pilot, and some are planned. Each one is labelled, with how far it has been tested, on Where We Stand.
How data is handled is on the Security and Data Handling page.
Not sure whether it fits your case?
Ask us directly. If EthosGuard is the wrong tool for what you need, we will say so.