The Human Signal Index maps narratives in the information environment. Because that work only matters if people can trust it, here is exactly how we operate — and how to hold us to it.
No finding is published, no account is restricted, and no enforcement happens automatically. A person reviews and decides every time.
Each claim links back to the source it came from. You can follow a published finding all the way to the original evidence.
Reviews, approvals, and changes are written to an audit log with who acted, when, and why — and they can be reversed.
We archive the underlying material and seal it with a hash so anyone can check it has not been altered.
A report stays a draft until a human reviewer approves it. We publish corrections — and the reason for them — when we get something wrong.
Access is limited by role, sessions are protected, and we run a vulnerability disclosure program so researchers can report problems safely.
You can ask what data we hold about you and request a correction. We do not sell your data.
We document how the platform works, which model versions produced a result, and where we have made mistakes.
Found a security flaw, or need to report misconduct? We have dedicated, confidential channels.