Consumer Reports creates new computer viruses

Kopji · 28th August 2006, 09:55 PM

Consumer Reports magazine creates over 5000 virus variants to test antivirus software:

http://redtape.msnbc.com/2006/08/con...ort.html#posts

Most of the comments are favorable and think it is a good idea.

Beleth · 28th August 2006, 10:06 PM

As long as none of them get out, it's not really a big deal.

Kopji · 28th August 2006, 11:38 PM

I have a bad feeling about this.

Heuristic sampling for viruses is a fairly invasive technique. More aggressive sampling of 'virus-like' activity would also have a higher rate of false positives.

The most pervasive viruses take advantage of known security holes, and the antivirus programs can make problems worse by going into endless detect/clean loops without the ability to actually prevent reinfection.

Kopji · 28th August 2006, 11:39 PM

Maybe they should work on that.

Wudang · 29th August 2006, 01:09 AM

That's interesting. Way way back I used to beta-test IBM's antivirus tool and they used to say that the important thing was to be able to detect variants of existing viruses "in the wild". It was my understanding at the time that there were a lot of viruses that were not in the wild - ie existed only in the labs - and that's what a lot of AV companies boasts were based on.

drkitten · 29th August 2006, 08:37 AM

Originally Posted by Kopji

I have a bad feeling about this.

I'm not sure that I understand your feeling, then.

Quote:

Heuristic sampling for viruses is a fairly invasive technique.

So? CU has never been known for their insistance on noninvasive evaluation techniques.

Quote:

More aggressive sampling of 'virus-like' activity would also have a higher rate of false positives.

And, again, this is in like with CU's other established policies. They test hard precisely because industry tends not to. And this is true regardless of whether they're testing anti-virus software or blenders.

28th August 2006, 09:55 PM	#1
Kopji Join Date: Apr 2003 Location: Mogollon Rim Posts: 7,697	Consumer Reports creates new computer viruses Consumer Reports magazine creates over 5000 virus variants to test antivirus software: http://redtape.msnbc.com/2006/08/con...ort.html#posts Most of the comments are favorable and think it is a good idea.

28th August 2006, 10:06 PM	#2
Beleth FAQ Creator Join Date: Dec 2002 Location: Not in a cave Posts: 4,134	As long as none of them get out, it's not really a big deal.
	__________________ Administrator and Head Moderator, The Skeptic's Guide to the Universe Forum Big Fan, Stop Sylvia Browne I will come back only after the words "Hi, Nyarl!" are returned to the post http://forums.randi.org/showpost.php...5&postcount=14 .

29th August 2006, 01:09 AM	#5
Wudang BOFH Join Date: Jun 2003 Location: Sheffield Posts: 8,245	That's interesting. Way way back I used to beta-test IBM's antivirus tool and they used to say that the important thing was to be able to detect variants of existing viruses "in the wild". It was my understanding at the time that there were a lot of viruses that were not in the wild - ie existed only in the labs - and that's what a lot of AV companies boasts were based on.
	__________________ Aphorism: Subjects most likely to be declared inappropriate for humor are the ones most in need of it. -epepke

28th August 2006, 11:38 PM	#3
Kopji Join Date: Apr 2003 Location: Mogollon Rim Posts: 7,697	I have a bad feeling about this. Heuristic sampling for viruses is a fairly invasive technique. More aggressive sampling of 'virus-like' activity would also have a higher rate of false positives. The most pervasive viruses take advantage of known security holes, and the antivirus programs can make problems worse by going into endless detect/clean loops without the ability to actually prevent reinfection.

28th August 2006, 11:39 PM	#4
Kopji Join Date: Apr 2003 Location: Mogollon Rim Posts: 7,697	Maybe they should work on that.

29th August 2006, 08:37 AM	#6
drkitten Penultimate Amazing Join Date: Mar 2004 Location: Wits' End Posts: 21,647	Originally Posted by Kopji I have a bad feeling about this. I'm not sure that I understand your feeling, then. Quote: Heuristic sampling for viruses is a fairly invasive technique. So? CU has never been known for their insistance on noninvasive evaluation techniques. Quote: More aggressive sampling of 'virus-like' activity would also have a higher rate of false positives. And, again, this is in like with CU's other established policies. They test hard precisely because industry tends not to. And this is true regardless of whether they're testing anti-virus software or blenders.