Discussion in 'other security issues & news' started by ronjor, Apr 19, 2019.
By: Rob Enderle | Updated April 18, 2019
HP Adds New Malware Protection Solution to Latest Laptops
Pretty cool stuff, I believe that Deep Instinct is comparable to Cylance. I hope in the future that more companies will start to offer these next gen AV's, that are hopefully less annoying and resource hungry.
It's still AI by any other name. It's still an AV dressed in a pretty frock.
Well, they claim it works differently when compared to a standard AV. In theory you almost don't need any updates and it doesn't need to be connected to the cloud. And it should not produce too many false positives. I take this any day over the bloated crap that is produced by Avast and BitDefender to name a few.
this is useless POS. marketing gimmick, nothing more.
And you have tried it personally as the basis for this remark?
Unfortunately we seen this rodeo before. They all cling to AI-Machine Learning etc now which is the current trending "marketing gimmick" for some that are no better than a Norton's or what have you. Maybe even not as effective. In my unbiased opinion of course This camp closed the door on AV's a long time ago and never looked back. Also not even a sliver of a malware issue either which raises my suspicions by leap n bounds.
Only time I ever got popped was running a freaking AV. Not a whisper or whimper since going dark from those and implementing Non-AV third party solutions.
Yes. I spent two years messing with a machine learning system for non av purposes. It's not bad, but didn't prove any better, then human knowledge. It's only is good as the training data, and even if they have a million samples to train with if the million and first is an outier good luck.
Secondly year ago I ran almost 900 samples every type of malware against my setup. The two av type solutions got around 85-90 percent which isn't bad. Appguard, HMPA and excubits products (MZwritescanner and pumpernickel) where 100% spot on. Just they take work.
Thanks to the nth for your expeience info, Peter! It's a strong indicator that AI probably isn't yet ready for prime time.
However, doesn't anyone here think that, conceptually at least, AI is the wave of the future -- not only as pertains to security but in many other ways, as well?
I wonder, will HAL 9000's ilk ever come to be? I expect so, & I'm kind of disappointed that I likely won't be around to see it.
I was an ardent fan of I. Asimov in my youth. I read & re-read everything fictional that he wrote. My favorite was "I Robot." Some day, the 3 laws of robotics may not seem so far-fetched.
And as @Peter2150 points out, pushing samples locally on your O/S (with a handy dandy image for restore just in case) is where the rubber really meets the road. It's the ultimate and absolute acid test where all your security programs light up with alerts-interruptions and otherwise captures that I seen many AV's choke completely out and indeed get overrun. Not the case with third party programs strategically implemented to signpost most any vector or factor the worse of the worse might try to bypass.
For my part AI will remain just that-artificial, whereas a good reliable time tested third-party set of a user's preference for security shielding is deliberate depending on what it's designed to do.
Also like @Peter2150 and perhaps others, we anymore have to funnel that malware crap locally (in testing) because if you have entry points covered like a solid granite block, it's doubtful even at best that any artfully crafted malware from past or modified today will have much success going any further than raising a flag and getting pinched at the start
AI is like we humans. It learns from experience, and can only learn from what it has experienced. Let me give you an example from my other life, working with dogs. 1) an animal that has only experienced love from humans. It will be trusting and want to meet every human it encounters. I know some dogs like that and they are a lot of fun. 2) and animal that grew up totally mistreated. He will be unfriendly and very wary of strangers. He will assume you mean him harm, and it can take a long time to earn his trust. Very satisfying when you finally do.
So do I think AI will be the wave of the future. Probably, but not because it's better, but because humans are getting lazier and don't want to have to think for themselves.
Quickly coming back to topic-Hewlett-Packard has been around throughout all the years PC Security is enjoyed getting something of a handle on matters concerning foulwares and their spawns and if they integrate a tighter control programming method that prevents it's products from getting bit that's a plus any way it shakes out.
But like anything else. the proof will be in the results. There's plenty of competition to go 'round but only the proven's grab attention from those serious about PC safety
It is feasible to expedite AI's learning curve by feeding it a massive, steady diet of malware. I read *somewhere* that that is what Cylance did.
That's true, but, if someone comes up with malware that's not in the diet, then your back at square one.
How about an (a) AI & (b) OSArmor & (c) HIPS & (d) signatures bundled together in one AV?
For my money you could skip a) and d). I am currently running this machine with no AV
AI is one factor to increase security. Just like heuristic and behavior detection are.
I also use no AV. Even so, if there were an AV with 4 components (AI, behavior blocker (BB), HIPS, & decent sigs), I would pay a reasonable price for a license, just to support its further development.
Bear in mind -- I think it is programmatically possible for an AI to learn more than simply spotting malware. It could also learn how best to deal with alerts resultant from BBs & HIPS that are set to be VERY highly aggressive. "Educating" an AI to be an "overseer" could help to prevent "average joe" from being overwhelmed by jillions of FPs & obscure alerts.
Had to give this some thought. Unfortunately I don't see it. First the "average joe" is to lazy and/or foolish. Example of foolish can be found on the MT forum. User put NVT's OSA on a family member's computer. Member didn't like the tray icon color but found a solution himself. He got it to turn white by disabling OSA. What can I say. Second thing is you have to constanly upgrade your education. For example the worst case of Powershell is a DLL that compiles directly into memory. No file ever hits the disk. How to deal with it was posted here on Wilders. Do you know what it is?
Here's my real problem with AI. I've been working with computers for almost 40 years. I've learned when there are problems of any kind only about 50% of them are solved with logical solutions. The rest end up being solved with off the wall illogical solutions, that can take some effort to find. AI by it's nature can only find the 50% the rest take human ingenuity. So I agree that AI is mainly a marketing thing..
Aloha Pete-sensei: I suggest we revisit this discussion in the year (say) 2030, if we are still extant. I believe that this discussion will be moot, on or before that year. It already is theoretically possible to build a computer that can be as teachable as a human, not only in a digital way, but in an analog way as well.
Yeah but can we build humans smart enough to how to use it.
The point is, you won't need to be smart to use it, any more than you need to be smart to use a toaster or a laundry machine.
Only one big difference. A toaster and laundry machine don't have a human behind them trying to defeat. Malware does.
"They claim", Avast, BitDefender and others claim quite a bit as well. Also standard/old AV's add new tech as well, so there isn't much difference between "next-gen" AV's and normal AV's.
The fact that it works without cloud connection is an interesting feature though. However, so far I haven't seen next-gen/AI based AV's score better in protection tests, and they have more FP's as well.
Separate names with a comma.