i can't remember if in 1995 we would write the code to be AT&T or if that came a bit later.
perhaps the scraper thought it was an HTML entity starting with the & followed by the T and it added the semi-colon.
or maybe it originally was posted in 1995 without the semi-colon and before it got indexed in 2002 by archive.org someone ran a regex style replacement on old static html files at netscape to "fix" html entities.
it just doesn't seem likely that a static html file from 1995 would have had AT&T; in it, it seems more likely that it was transformed either by the scraper or some internal script run after it was originally posted.
i found an old email version of the press release on Lunds University website that does not have the semi-colon. http://www2.ldc.lu.se/temadag95/javascript.txt
and what looks like an archive of the 1995 version of the java website hosted at University of Oviedo: http://www6.uniovi.es/java-http/pr951204-03.html