It's been years since desktop signature-based antivirus provided a significant improvement to security. Every time there's a cryptolocker outbreak, I see people scrambling to make decisions like "we need to replace McAfee with Kaspersky", as though they feel that's their answer.

When you try telling an insurance auditor "we have a whitelisting application, nothing runs unless I've approved it, products like Symantec Endpoint are unnecessary in that environment", you first get a confused look, then you hear "ok, so you're DON'T meet the minimum basic security requirements, let me write that down".

It's gotten to a point that it's actively part of Intel Security's advertising, with a recent partner promotion pushing to "help your clients meet their compliancy requirements". The brochure never even mentioned actually securing anything, just how it ticked various boxes.

Unless the compliance auditors are happy with the software just being installed and don't check whether it actually runs. That would still be a colossal waste of money for the licenses but at least it would not compromise your security.

All operating systems that have some way to allow people to run malware, will get malware. Windows, OSX, GNU, Android all can get infected quite easily. Then there's iOS where you cannot, and instead Apple decides which software you can or cannot run.

The downside is of course that you cannot run any software going against the corporate values of Apple.

If you want the right to shoot yourself in the foot, AV is the necessary evil you must have, unless of course you're sure you'll never visit a website that contains an exploit, old or zeroday, against you browser or its components, and you will never open a office document, PDF or executable that has malware in it. And even then you can get owned.

No. AVs are actually pretty useless at stopping anything except the most basic attacks (and sometimes, not even that - just look at Cryptolocker).

Use Google Chrome (really! Firefox isn't even playing in the same league security-wise), disable Flash player, only run trusted executables with valid digital signatures.

Can you elaborate just a little on this please?

In Firefox all tabs run in the same process and thus inherently can't be sandboxed (because it needs to write to the disk cache and save files the user downloads)

Uninstalling Flash, Adobe reader, Office and JRE, and using Chrome with adblock also helps you enormously, but is still a far cry for any user having difficulties with finding the download-button from sourceforge.

Getting a signing cert is easy as just buying one from Honest Achmed's Used Cars and Certificates, so the only real use for signed software with malware protection is to manually maintain your own list of trusted signers.

Of course they do well there – the vendors use those as a primary marketing feature. It's like learning that Oracle does well at a TPC benchmark they'll be printing on glossy brochures.

The question a buyer should be asking is “What percentage of attacks the average Internet user faces are stopped by this product?” and that has been declining steadily since the 90s because virus authors can easily test before releasing a new version and confirm that they've managed to avoid the current signatures. It doesn't matter that your product is great at stopping last year's malware if that's not what exfiltrates or encrypts your data.

> Uninstalling Flash, Adobe reader, Office and JRE, and using Chrome with adblock also helps you enormously, but is still a far cry for any user having difficulties with finding the download-button from sourceforge.

The part that you left out is that using Chrome gets you all of those but ad-blocking. It's true that it's hard for many users to operate securely but millions of them have managed to install Chrome and that's far more effective than any security product on the market.

I honestly cannot imagine a better way to objectively test how well the products fare against attacks against an average Internet user.

Edit: If I was not clear, nobody tests with historical samples anymore. Only live attacks are being used for tests.

1. Malware author releases something new

2. Users start getting compromised

3. Antivirus vendors start getting samples and analyzing them

4. New signatures are released

5. Clients download and install the new signatures

That cycle used to work better but in the Internet era it's a given that malware vendors are taking advantage of the substantial time delays between steps 4 and 5, which are often measured in hours or even days, and will change their code as soon as new signatures are released.

When someone reports results and they specify that the percentages are based on a historical library, that tells you little about what it'll do for you now. When they tell you that results are based on samples collected in the month prior to the test, which is what AV Test and AV Comparatives say they do, that's less stale but since it's starting after the vendors have already completed the entire process it still doesn't tell you how long you'll be exposed between steps 1 and 5 or whether some malware authors are consistently staying ahead of the loop.

This is really coming back to security fundamentals: trying to enumerate all of the bad things on the internet is futile. The better strategy is removing the ability to run programs which aren't on a known-good list but that breaks a lot of legacy practice.

> I honestly cannot imagine a better way to objectively test how well the products fare against attacks against an average Internet user.

The most reliable way to do this would be to simulate randomly surfing around the web, being sure to click on all of the ads, while monitoring for changes to existing programs or new programs, access to files the browser had no reason to open, and unexpected network connections.

Unfortunately, not even this approach will work. No. To be totally safe, you have to whitelist by digest of the exe and command-line arguments. Which basically means that you have to know the the OS works internally.

This is basically what Apple shipped in OS X 10.11 where you can trust third-party developers but System Integrity Protection (https://support.apple.com/en-us/HT204899) tries to limit the damage that even getting root can cause.

I asked a netadmin friend for some advice, and he said "Oh, that thing is just there to tick boxes - no techie would ever use it. 'Does it have a GUI configurator?' -> box ticked". Meanwhile the techies just configured the modem like any other bit of cisco equipment: via the console.

(of course I've no doubt there are plenty of people smart enough to write their own but that's not the concern)

I just don't see any accidental options.

See also:

https://news.ycombinator.com/item?id=8257250 (DrawBridge & commentary)

Oh and the follow-up in the comments: (Regarding Sandboxie reconfiguration mitigating some of the attacks)

> I tested with SBIE4 and indeed it is quite different and has security improvements over what we tested in our report. Great job by the author!

> However, our kernel exploits still work flawlessly – no issues there. You need to remember that SBIE3/4 (and other similar app sandboxes) are kernel mode drivers and they will always have some fundamental limitations. Blocking access to specific DLL’s is far from a practical use case as you have no idea where the next patch is going to come (BTW similar capability has been available in HIPS technology for several years and very few people use it)

> Once time permits, we’ll probably update our analysis with the latest SBIE4. However, this is not our core focus, the idea of this research was to simply enumerate the fundamental limitations of application sandboxing tech.

If you've read this far, you might also be interested in other Bromium research: https://labs.bromium.com/category/research/

Of course, folks might be first in line to point out that nothing's perfect, and user-error can get around a lot of security... Nobody would argue sandboxing is an excuse to not patch systems, except with antivirus you might increase your attack surface while trying to detect attacks with something that can't be easily circumvented. Trade-offs. Can you trust sandboxing to prevent damage from malware or do you need some kind of warning that something strange is going on?

    Anti-Virus software is a trash-fire.

I've long argued that the only sustainable security strategy is education.

If you're a developer/engineer/consultant/rockstar/ninja/etc. and never bothered to learn how to write secure software, start here: https://paragonie.com/blog/2015/08/gentle-introduction-appli...

And, even for those of us who are somewhat savvy, having refreshers to remind you things like never, ever, click on a link in an email, navigate through the website instead. Be hyper cautious about opening every attachment, period. Ensure that your system firewall is set to default deny, and think twice about opening up rules when an application demands them - and consider making them temporary rules. (Little snitch is great for that). Make sure the OS is updated routinely....

With that said, operating systems could help us out a little by reducing the almost infinite number of threat surfaces that exist so that we can more easily audit our system. The sheer number of places that an auto-launching/malware/kernel extension can hide in OS X makes it next to impossible for me to figure out whether my system has been compromised - particularly if something is able to hide itself from Little Snitch.

And I won't even go into the insane number of network accesses that most applications want these days...

But yes, the only sustainable security strategy is education.

I think that's an incorrect standpoint to take when you factor in how technologically agnostic most folks are. I would better think of it as "security is default, but you can disable it if you really, really want it that way."

There is a good amount of intuition that goes into it. Such vulns are relatively rare compared to the things they run into multiple times a day.

Attackers agree. Even outside of the context of remotely exploitable ring-0 vulnerabilities that require no user interaction, they provide people with a false sense of security.

> Such vulns are relatively rare compared to the things they run into multiple times a day.

That's because criminals are fat and happy with lower hanging fruit. Why bother exploiting AV software when you can just trick people and the AV software is already ineffective at stopping e.g. ransomware?

They apparently use their own product on their email server, which unpacked the POC by guessing the password of the archive, scanned the uncompressed file and triggered the bug that was being reported. Love it !

Increasingly, my non-computer savvy family members ask me what kind of anti virus they should use. I used to pick one to tell them since I know they aren't as cautious as I am, but I am not sure I have a good answer for them any more. Has AV software reached the point that a lay user is more vulnerable with it than without it?

My other recommendation is to use a tablet for things like online banking. (Yes, even an outdated Android tablet is probably less likely to catch malware that will steal your money than an average computer.)

The product is geared towards medium-to-large networks so it's a little hard to find the prosumer $20/yr plan. Here's a link if anyone else is interested: https://www.opendns.com/enterprise-security/threat-enforceme...

Is this a real thing? Hands up, who runs Norton on Linux? Is it because it is used as a central back-end / service to check attachments. But then why does it run as root?

The phrase "commonly affected" is the place to make an argument here, but I'm sure people take the easy option of just running an antivirus.

When my company gets asked why we answer "no" to that question, my canned response is "because anti-virus software would almost certainly be the most exploitable vector on our systems".

It looks like the researcher sent a proof of concept zip file to symantec which was pw protected with a common password. Symantec's system then tried the common password, extracted the zip, scanned the POC code inside, which crashed their own system.

From the report: Project Member Comment 1 by taviso@google.com, Yesterday (42 hours ago)

I think Symantec's mail server guessed the password "infected" and crashed (this password is commonly used among antivirus vendors to exchange samples), because they asked if they had missed a report I sent.

They had missed the report, so I sent it again with a randomly generated password.

I understand how this vulnerability can be used to corrupt the heap, as it's writing more data than malloc was asked to reserve, so it can overwrite memory allocations from other parts of the program.

I am curious as to how would one create a reliable remote code execution exploit out of this? I guess that one may be able to find a function pointer somewhere to overwrite, and use that to control program flow to your shellcode - but as this is dynamically allocated memory, could it not be adjacent to pretty much anything?

How would an attacker approach making a remote code execution exploit, given these constraints? Is it possible in practice or more theoretical?

(I'm not challenging this classification, just would really like to know how this works!)

That rings a bell -- I remember back in the early/mid 2000s, when the AV vendors started to port their products to Mac OS X. The darwin (OSX) kernel/driver mailing lists seemed to get a lot of questions from AV devs, asking how to do things in the kernel that really, really, really should not be in the kernel. It was at that point I resolve to never run any AV software.

This is just a stupid, lazy way of doing business.