Every upcoming submission deadline for computer science and AI conferences, rebuilt daily from two community-maintained lists published as versioned files on GitHub. Each entry links straight to the conference's own page, shows how long is left, and says whether the date has moved since ORBIT started watching it. The page is free, needs no account, and carries research and scholarly-publishing news alongside the deadlines.
The two sources are community-maintained lists of computer science conferences: the CCF deadline list and Hugging Face's AI deadline list, both MIT licensed. Between them they cover artificial intelligence, computer graphics and multimedia, computing theory, databases, distributed and parallel systems, human-computer interaction, networking, security, and software engineering.
There are no humanities, life-sciences or general engineering conferences here, and ORBIT would rather say so than let you assume otherwise. If you are looking for a journal rather than a conference, ORBIT's journal index covers roughly 98,000 of them across every field, and a suggestion is made from your abstract alone.
Conference deadlines are extended, and a list that quietly rewrites a date tells you nothing about how stable it is. ORBIT records the deadline it read on each daily check and compares it against the one already stored. When they differ, it keeps the old date and stamps the moment it changed, so an entry can say that a deadline moved three days ago and what it was before.
The wording is deliberately careful. When nothing has moved, the page says the date is unchanged since ORBIT started watching it — not that it has never changed, which is a claim about a period before we were looking and one nobody here can honestly make.
This part is ORBIT's own. A daily job re-reads the author-guidelines pages of the journals ORBIT can run a readiness check against, hashes each one, and records when the page changes. Nobody else publishes this, and it is the reason a submission deadline page belongs on this site rather than anywhere else.
The entry says only that the page changed and links to it. It does not tell you what changed, because a machine comparing two hashes knows that something moved and not what it means — and a wrong summary of a journal's submission rules is worse than no summary at all.
ORBIT hosts none of it. The conference lists are public files in public git repositories, and the news comes from the RSS feeds Nature, Science, Retraction Watch, The Scholarly Kitchen and DOAJ publish for exactly this purpose. ORBIT stores a title, a date, a short summary and a link, and sends you to the original.
Two obvious sources are deliberately absent. LinkedIn's User Agreement prohibits scraping outright. Reddit's free API tier is licensed for non-commercial use only, and ORBIT sells credits. Both were ruled out on their terms rather than on their usefulness, which is the same reasoning that keeps journal impact quartiles off ORBIT's journal suggestions.