AI Disclosure

There is no AI conversation feature

LumenUs does not offer an AI chatbot, an AI companion, or any AI conversation feature. Nothing on this website talks with you, listens to you, or generates responses to you. The words on this site were written and reviewed by people, under our Editorial Standard. No conversation with you is collected, processed, or stored, because no conversation feature exists.

The first: reading your papers

In the LumenUs app, you can photograph or upload papers such as a death certificate, a W-2, a bank statement, or a bill. Each page is read by Claude, a model made by Anthropic, which returns the details printed on it. The app says so on the screen where you add papers. What it reads is shown to you beside your photo, and nothing is added until you confirm it. It can be wrong, so you check each detail against your paper.

• Anthropic does not train its models on your papers, and deletes each page and its reply within 30 days, unless the law requires it to keep them longer.
• We keep no copy of your photos. They stay on your own device until you have gone through what was found.
• The reader is instructed never to return the cause or manner of a death or a parent's name. A Social Security number is read only from a death certificate, and is stored only if you confirm its last four digits.
• The reader does not converse with you, give advice, or decide anything for you.

The full details are in our Privacy Policy, Section 5.

The second: reading letters before they reach a family

A member can share a page that remembers the person who died, and anyone with its link can write to the family. Before a letter reaches the family, Claude, a model made by Anthropic, reads it for one thing: to hold back the few letters that would hurt or exploit the family, such as cruelty, threats, requests for money, or advertising. The page tells the writer this before they send.

• When the model is unsure, the letter goes through. A letter it holds back never reaches the family, and we delete it after 30 days.
• It reads the letter, never the person: it does not judge or assess the writer, and it does not look for signs of crisis.
• The writer's email address is never sent to it.
• Anthropic does not train its models on letters, and deletes each letter and its reply within 30 days, unless the law requires it to keep them longer.
• It can be wrong, and the family can remove any letter at any time.

The full details are in our Privacy Policy, Section 5.

Our commitments for any use of AI

These commitments govern the paper reader and the letter reader, and they will govern any other feature that uses artificial intelligence. This page will describe any such feature plainly before it is available to anyone:

• You will always know. We will disclose clearly and conspicuously, wherever anyone could reasonably mistake software for a person, that they are interacting with software.
• It will never simulate the person who died. No LumenUs feature will ever impersonate, simulate, or speak as a person who died, or as any real person, living or dead. That is a line we will not cross.
• It will never claim to have grieved. Software has no lived experience of loss, and nothing we build will pretend otherwise.
• It will not be therapy. No AI feature will diagnose, screen, or treat any condition, and none will be presented as a clinician or as a substitute for one.
• Crisis resources stay free and ungated. No feature, AI or otherwise, will ever place crisis resources behind a login, a paywall, or a tier.

Crisis

LumenUs is not an emergency or crisis service. Crisis resources are published at lumenus.life/crisis-support and are always free and never gated: the 988 Suicide and Crisis Lifeline (call or text 988), the Crisis Text Line (text HOME to 741741), and 911 for an immediate emergency.

Contact

For questions about this page, contact hello@lumenus.life. This inbox is not monitored for emergencies. If you are in danger, use 988 or 911.