Privacy policy
Last updated: 14 October 2026
The Observatory is a small news service about artificial intelligence and Ethiopia. Mikael Alemu Gorsky runs it from Herzliya, Israel, and he is responsible for the personal information it handles. This page explains what information the Observatory receives, what it does with that information, which other companies handle it, how long it is kept, and how you can have it changed or removed. Please send any question about privacy to policy@mgorsky.net.
What the Observatory receives
When you subscribe on this website, the Observatory receives your email address, the publications you choose, the hour and the language you choose for each of them, whether you want them all in one email a day or a week, and the time zone that your browser reports. It then sends 1 email to confirm the address, and nothing else is sent until you follow the link in that email. If you do not follow it within 48 hours, the request lapses. The Observatory keeps a record of each request, confirmed or not, and it limits how many requests it accepts in a day from any one address.
When you change or stop a subscription through the link in an issue, the Observatory receives your new choices or, when you stop, the reason you give and any note you add.
When you subscribe, ask a question or write to the Observatory by email or through its Telegram bot, the Observatory receives your message, your email address or your Telegram account with the name that Telegram shows, and the time of the message. For email, it also receives Google’s check of whether the message really came from the sender’s domain, and it uses that check so that it never answers a forged address.
When you send a letter through this website, the Observatory receives the text of the letter and, if you give it, your email address.
When you visit this website, you are not asked for anything, and the site sets no cookies. It counts visits with GoatCounter, a statistics service that uses no cookies and keeps no lasting identifier for a visitor. For each page that is opened, GoatCounter records the page, the site or link from which the visitor came, and the visitor’s approximate country, browser, operating system, screen size and language. The private key in a confirmation or unsubscribe link is removed from the address before the page is counted. Visits by AI agents and crawlers, programs that name themselves in each request, are counted apart: for each such request the site records the program’s name, the page and the time, and nothing else, and these records are deleted after 30 days. The front page shows the Observatory’s latest videos: their pictures are loaded from YouTube, and a video’s player is loaded from YouTube (youtube-nocookie.com) only when you press play, after which YouTube may store information on your device under Google’s privacy policy. The pages load their fonts from Google Fonts, so Google receives the network address of each visitor’s computer, as it does for the videos’ pictures, and Netlify, which hosts the site, keeps ordinary records of the requests it serves. No third party serves advertising or other content on this website or in the Observatory’s publications.
Most of what the Observatory reads is public material: news articles, posts, newsletters, research papers, podcasts and videos. That material often names people and reports what they said. The Observatory keeps short summaries of it, written in its own words, each with a link to the original. This is the ordinary work of a news service and does not concern the Observatory’s readers.
What the Observatory does with it
The Observatory uses your address only to send the publications you chose, at the hours and in the languages you chose. It records each attempt to deliver an issue, so that no issue reaches you twice and a failed delivery can be tried again.
Questions sent by email or through the Telegram bot are read by the Observatory’s software, which looks for an answer among the notes the Observatory has already published and writes a reply that cites those notes. The reply goes back by the same route by which the question came. Letters sent through this website are kept in the Observatory’s archive of correspondence and are read by Mikael Alemu Gorsky; software does not answer them. Mikael Alemu Gorsky also receives a daily summary of what readers have sent, including the reasons given for stopping a subscription.
The Observatory limits how many questions one person may send in a day, and it keeps lists of addresses it will not serve, so that the service cannot be abused.
The Observatory does not sell, rent or lend personal information, does not use it for advertising, and does not build profiles of its readers.
Artificial intelligence
The Observatory’s software uses AI models to read and answer questions sent by email or Telegram, and to translate issues into Russian and Amharic. It reaches the models through OpenRouter, a service that passes each request to one of the companies that run the chosen model. The text of your question is therefore processed by OpenRouter and by that company, which may be in the United States, China or another country, and which keeps and uses the text under its own rules. The Observatory gives the models the text of your message but not your email address or your Telegram account. Letters sent through this website are not given to the models.
Other companies that handle the information
Several companies handle information on the Observatory’s behalf, each under its own privacy policy. Google delivers the Observatory’s email through Gmail, announces new mail through Google Cloud, hosts the Observatory’s videos on YouTube, from which the front page shows them, and serves the fonts of this website. Telegram delivers messages to and from the Observatory’s bot and channels. Supabase receives the forms sent from this website; the Observatory collects each form within minutes and then deletes it from Supabase. Supabase also keeps the count of AI agents’ visits described above. Netlify hosts this website, and GoatCounter counts its visits. OpenRouter and the companies that run AI models process the text of questions, as described above. LinkedIn shows the Observatory’s posts on LinkedIn, but the Observatory gives it no information about its readers.
The Observatory also publishes a read-only view of its working screens. That view hides readers’ names, addresses and messages.
The Observatory on YouTube
The Observatory publishes some of its issues as videos on its own YouTube channels. It does this through YouTube API Services, which Google provides. Google’s Privacy Policy, at https://policies.google.com/privacy, explains how Google handles information on YouTube, and the YouTube Terms of Service, at https://www.youtube.com/t/terms, govern the use of YouTube.
The Observatory uses YouTube API Services only to upload its own videos to its own channels and to manage those videos. For this purpose, the owner of the channels has authorized the Observatory’s software to act on them. The software keeps that authorization on the Observatory’s computer, together with the identifier, title, description, visibility and upload status of each video it has published. It does not read or store any information about the people who watch the videos, comment on them or subscribe to the channels, and it does not use YouTube API Services to reach any other account. The Observatory shares none of this information with anyone other than Google.
The authorization can be withdrawn at any time on Google’s security settings page, at https://security.google.com/settings/security/permissions. When the authorization is withdrawn, the Observatory stops using it at once and deletes the YouTube information it stores within 7 days. Questions or complaints about the Observatory’s use of YouTube API Services can be sent to policy@mgorsky.net.
Where the information is kept and for how long
The Observatory keeps its information in its own databases on a computer it runs in Israel. A copy of the databases is made every day, and the last 14 copies are kept.
Your subscription is kept while it is active. When you stop it, the Observatory stops sending at once and keeps only the record that your address has stopped, with any reason you gave, so that it does not write to you again. Records of subscription requests, including requests that were never confirmed, are kept with your address unless you ask for them to be deleted.
Letters, questions and the answers to them are kept in the Observatory’s archive of correspondence without a fixed end date, because they are part of its record. If you ask, the Observatory deletes your letters and questions together with your address.
The YouTube information described above is kept for as long as the authorization remains in force.
Your choices and rights
Every issue sent by email contains a link that stops the subscription, and you can also stop it by writing to the Observatory by email or on Telegram.
You may ask the Observatory what information it holds about you, ask it to correct that information, or ask it to delete it, by writing to policy@mgorsky.net. The Observatory acts on such a request within 7 days. Deleted information leaves the daily copies as they are replaced, which takes no more than 14 days. If you believe the Observatory has mishandled your information, you may also complain to the data protection authority of the country where you live, which in Israel is the Privacy Protection Authority.
Children
The Observatory is not meant for children under 16, and it does not knowingly collect information about them.
Changes to this policy
When this policy changes, the new version will appear on this page with a new date. If a change affects information you have already given, the Observatory will tell its subscribers before the change takes effect.
The Observatory’s terms of use set out the conditions for using its website and publications.