The standards change when someone shows where they are wrong or where a system broke them. This page says what to send and where it goes: one inbox, hello@ethotechnics.org, or the public repository.
Send postmortems, rulings, and inquiry findings from systems you have worked on or followed. The casebook scores failures a court, an inquiry, or a regulator established; the mechanisms catalog needs evidence of fixes working and failing.
Send a field report or case
A postmortem, research memo, or public record showing an automated decision system whose safeguards held or failed. A case established by a court, an inquiry, or a regulator can be scored in the casebook.
Say what the system decided, and for whom.
Attach the records you can share: decision logs, notices, rulings.
Say what changed for the people affected, if anything did.
Reviews are paid work, done by Kanav Jain, who wrote this framework, on healthcare AI systems: a safeguards review, a readiness sprint, investor diligence, or a clinical AI safety evaluation. Lessons from a review are published only with your agreement, and anonymized.
Say what is wrong, unclear, or missing: a broken link, a clause that cannot be tested, an example you needed and could not find. A substantive critique gets a written answer saying what changed, or why nothing did.
Feedback inboxes
Send notes, critiques, or missing references
Share unclear steps, missing citations, or examples you want us to add.
To object to a clause of a standard, use the Object link beside it in the
standard's clause register. Each objection is a public GitHub issue,
answered within 30 days, and the site's own record lists
every answer.
This inbox is for fixing the site's content, not for challenges to
decisions. To challenge a decision an institution made about you, use that
institution's appeal route. The
self-defense checks list what
you can ask for and the wording to use.