Instead of matching valid sequences, match invalid characters. Then use $-[0] to find the position of that match. (The @- array is documented in the "perlvar" manual page.)

use strict;
use warnings;

while (my $sequence = <DATA>) {
    chomp $sequence;
    if ($sequence =~ /[^ATCG]/){
        warn "Sequence '$sequence' has invalid character after " . $-[
+0];
    }
    else {
        print "Valid sequence: '$sequence'\n";
    }
}

__DATA__
TAAGAACAATAAGAACAA
TAAGAACAATAAUAACAA
TAAGAACAATAAGAACAA
[download]

You don't need to split the sequence up into individual characters and process each one separately. That's slow.