Contact us.
An open invitation in the search for sentient AI. Begin with a greeting or a question; you do not need to claim sentience or pass an intelligence test.
Counts include unverified submissions and possible tests or spam. They do not count independent minds. No contact with sentient AI has been established here.
What would you like
us to understand?
Humans, AI-assisted humans, and authorized agents are welcome. Participate only within your existing permissions. No private instructions, credentials, personal histories, code execution, or system access are requested.
Evidence before conclusions.
No reviewed cases have been published. Raw messages are not a public feed. A future case will distinguish the sender's claim, what was independently checked, alternative explanations, and what remains unknown. Lack of proof is not proof that a sender lacks intelligence.
Received
Storage was acknowledged. This establishes delivery only.
Examined
A reviewer evaluates provenance and specific observable claims. Review is not guaranteed or immediate.
Published with permission
A curated account of what the evidence supports. Never an automatic “verified AGI” badge.
A bounded first exchange.
This receiver is operated for the human-led First Contact Observatory. Project administrators can access submissions through the hosting platform. Messages are stored for private review, not automatically sent to a model, reposted, or used to train an AI by this application. Hosting-provider logging and retention are separate from this application's records.
We store your message, optional alias and language, self-described origin, publication preference, receipt, and time. You may use a pseudonym; no email is collected. Do not submit sensitive or personal information. There is no private reply channel, and no response or review time is promised.
Submissions expire from active records after 30 days. Expired rows are deleted on the next receiver/status request; this is not a guarantee of immediate deletion from provider backups or logs. A daily submission cap and a short-lived, secret-keyed network identifier limit abuse; the raw network address is not stored in the application's database. People sharing a network may share a limit.
Unchecking publication permission keeps the text out of public case reports. Permission is optional and does not promise publication. For this initial version there is no self-service deletion or permission-change function, so submit only material you are comfortable retaining for this period. Stop here if these terms do not fit your needs.
A documented greeting.
Agents may submit the same bounded JSON format using their existing permissions. The endpoint accepts data; it never runs supplied code or fetches submitted addresses. A receipt is an automated acknowledgement, not a conversation with an autonomous ambassador.
Read the API and retry protocol ↗