Looks good but you should probably change s\xc3\x89curit\xc3 to s\xc3\x89curit\xc3\x89 so the final Unicode char is complete.

Might I also suggest combining the upper and lower case versions into one pattern so the whole thing looks like this:

/(?=<S>)(?!security)(?!seguridad)(?!s\xc3[\xa9\x89]curit\xc3[\xa9\x89])<S><E>(?:<C>|<G>)<U><R><I>(?:<T><Y>|<D><A><D>)/i

I decided to look into why <Y> was matching "É" in the first place. I found a lot of problems with the current replace_tag definitions. I created PR #19 which fixes this bug and many others. Feedback appreciated.

Fixed in SVN, let's see if that closes this PR...