Trekkie said:
two non-propigating trojans (one which is even debatable if it's a 'trojan' or not) this is chump change.
The problem is that this could easily be made into a propagating problem combine *with* some of the other vulnerabilities out there.
Here is a question: You get a small DMG from a friend who is in your address book. You presumably trust them. The email says, "I found this great utility to control your internet preferences, to avoid a trojan that hijacks your help viewer. This was just announced today, see <inforword article or whatever>." What do you do? 90% of the people will click it.
What happens to you? It is the trojan itself. Instead of the email, it is the trojan mailing itself out to everyone in your address book - something can EASILY do that. Then it can easily do an rm -rf /.
Or it tries to start up the remote desktop self-start (from the command line there is a command for it).
Or it uses the command line to open up a small additional application on the disk image which can do more.
Or it uses wget (or similar) to download and start something at the command line.
Additional possibilities depending on what it does above and what is on your system:
Or it can open up ports on your firewall or just turn it off.
Or it can install a key stroke recorder and periodically send out "interesting" keystrokes (e.g. anything typed will connected using an SSL connection).
Or it can say "Disk Utility needs your admin password to mount this signed image"? Most people would type it in because OS X has us (I might do it) conditioned to do so. Then it could email (or send via TCP without using your mail program) the admin plus your IP somewhere.
Or it can install a background application to listen on a particular port while re-configuring your firewall to allow that port to be open.
Or it can email documents in your "Documents" folder somewhere. (e.g. just email files < 5K in size. What kind of info is in those files?)
Just because the demo is non-malicious and doesn't exploit all the possibilities doesn't mean they are not there. There are a tons more things it could do from the really subtle so you would never know it to the really damaging and everything in between.
There are enough vulnerabilities out there that it could be done. Worms are easy to write. Viruses are too. Trojans are too. Someone with a good knowledge of Unix could write one to take advantage of 8 to 10 of the most common open vulnerabilities and it would be a huge problem.
Someone who wrote one that exploited 8-10 Unix problems (including Mac and Linux) *and* 8-10 Windows problems would cause complete havoc because there wouldn't be only one method of propagation to stop, there would be multiple routes to infection for just about all of the machines on the net. And if Unix machines are infected, Windows are, and Macs are (and perhaps Cisco router vulnerabilities), how the heck will people get updates to fix them? CD? It could be done and it will be done, it is just a question of when. Up until now you've had amateurs doing it, script kiddies etc.
Hopefully it will remain that way, but I believe at some point there will be someone who decides to exploit many (not just 1 or 2) problems at once, who doesn't make stupid programming errors (e.g. like the Morris Worm or many others), who is based where they don't care (e.g. North Korea or some small, but big-time criminal organization or terrorists) and has a specific malignant purpose. Then many billions of hours of work will be lost or stolen.