There are two ways to do this. The first is to use an ordinary match and then negate the result:

if (!($string =~ /keyword/)) { print "match failed"; }
[download]

or:

unless ($string =~ /keyword/) { print "match failed"; }
[download]

The second is to use the negated binding operator !~:

if ($string !~ /keyword/) { print "match failed"; }
[download]

See perlop#Binding-Operators.

Hope that helps,

perl -E "my $string = 'abc efg hijk lmnop'; if($string !=~ m/defg/) {s
+ay $string;}"
abc efg hijk lmnop
[download]

If it does the job without side-effects, it's probably correct!

But is there some reason you couldn't have answered this yourself... or at least satisfied yourself that your code does or does not work?

Update on my oops: TY, Athanasius who pointed out my typo and, implicitly, that -E does not invoke warnings, whose use would have also pointed out my mistook. With a

use warnings;<c> added to the one-liner this is evident:</p>

<c>!=~ should be !~ at -e line 1.
Use of uninitialized value $_ in pattern match (m//) at -e line 1.
Argument "abc efg hijk lmnop" isn't numeric in numeric ne (!=) at -e l
+ine 1.
[download]

With the correction, and a more generous example of strings, however, IJW:

C:\>perl -E "use warnings; my @strings = ('12345 defg', 'abc efg hijk 
+lmnop', 'xyz defg abc'); for my $string(@strings) { if($string !~ m/d
+efg/) {say $string;}}"
abc efg hijk lmnop
[download]

Which just goes to say, OP had the "not much wrong" idea.

AB CDE         #don't know number of space between
AB     CDE     #5 space  =~ /AB.....CDE/
AB       CDE   #7 space  =~ /AB.......CDE/
[download]

You could use either of those:

/AB +CDE/; # one or more spaces
/AB\s+CDE/; # one or more spacing characters (incl. spaces, tabs, end 
+of line, etc.)
[download]

If it's just spaces, then you can use

/AB\s*CDE/
[download]

But if they can be any characters except the sequence CDE, you can use

/AB(?:(?!CDE).)*CDE/s
[download]

(?:(?!PAT).)* is to PAT as [^CHAR]* is to CHAR.