my $str="abcd";
my $letters = "ab";

print not_in($str, $letters),"\n";

sub not_in {
 my $str = shift;
 my $letters = shift;
 my (%hash1, %hash2);
 my @arr1 = split //, $str;
 @hash1{@arr1} = @arr1;
 my @arr2 = split //, $letters;
 my @deleted = grep $_, delete @hash1{@arr2};
 @hash2{@arr2} = undef;
 delete @hash2{@deleted};
 return if %hash2;
 join '', keys %hash1;
}
[download]

Updated to more accurately reflect the requirements :-)
Actually, this one doesn't remove characters one for one as it seems you want to. If a character appears in the second arg, it removes all of that character in the first arg. That's not what you wanted is it??