The first "announcement" of the Green&ndash;Tao theorem on arbitrarily long arithmetic progressions of primes was the appearance of their preprint on the arXiv.  When I saw that preprint, I posted an article to the USENET newsgroup sci.math, somewhat incredulously asking whether this was the first public announcement, and <a href="https://groups.google.com/d/msg/sci.math.research/r4xXHsqmMuE/FRjFafZV8ccJ">Tao replied</a>:

> It is our first public announcement, yes.  Given the track record for announcements for well-known conjectures in number theory, it seems a low key approach is appropriate. :-)