SE Labs

Posts filed under 'AI'

Predictably Evil

pmr-1176337

A common criticism of computer security products is that they can only protect against known threats. When new attacks are detected and analysed security companies produce updates based on this new knowledge. It’s a reactive approach that can provide attackers with a significant window of opportunity.

It’s why anti-virus has been declared dead on more than one occasion.

Latest report now online.

Security companies have, for some years, developed advanced detection systems, often labelled as using ‘AI’, ‘machine learning’ or some other technical-sounding term. The basic idea is that past threats are analysed in deep ways to identify what future threats might look like. Ideally the result will be a product that can detect potentially bad files or behaviour before the attack is successful.

(We wrote a basic primer to understanding machine learning a couple of years ago.)

So does this AI stuff really work? Is it possible to predict new types of evil software? Certainly investors in tech companies believe so, piling hundreds of millions of funding dollars into new start-ups in the cyber defence field.

We prefer lab work to Silicon Valley speculation, though, and built a test designed to challenge the often magical claims made by ‘next-gen’ anti-malware companies.

With support from Cylance, we took four of its AI models and exposed them to threats that were seen in well-publicised attacks (e.g. WannaCry; Petya) months and even years later than the training that created the models.

It’s the equivalent of sending an old product forward in time and seeing how well it works with future threats. To find out how the Cylance AI models fared, and to discover more about how we tested, please download our report for free from our website.

Follow us on Twitter and/ or Facebook to receive updates and future reports.

Review: ImmuniWeb On-Demand Application Security Testing

hacked-1764593

What do a start-up, small business and enterprise have in common?

They all have one or more websites.

That’s not a very humorous punchline, but the security implications of managing business websites aren’t funny either.

In an age when extremely large organisations are being hacked, as well as specialist security companies, website security could not be a more serious business. Throw into the mix regulations such as the data protection act and the incoming GDPR legislation and being the person responsible for the company website just became positively horrible.

A website is a business’ public face, whether it be a local taxi company or a global pharmaceutical giant. It is virtually impossible to do business these days without a website and maintain credibility, but a website hack instantly harms any company’s standing.

How do websites get hacked? Sometimes the attackers will focus on compromising the site’s administrator, but more often than not (in our experience) the site itself is attacked directly by means of an exploit.

Such an exploit could be a aimed at a vulnerability in the platform, such as WordPress, or the server’s operating system. Sometimes the hosting company itself is targeted: a good value-for-money proposition for an attacker who wants to run one attack and gain access to thousands of websites.

Will AI save our sites?
Artificial intelligence is great but people are often necessary for some tasks. ImmuniWeb understands that. Assessing the security of a website is non-trivial and, while automated tools exist to test for the presence of various vulnerabilities, often it takes a human brain to really get to the bottom of a problem. Much in the same way that SE Labs uses people to enhance security testing, ImmuniWeb adds the personal touch to checking the quality of a website’s security.
The service provides testing for vulnerabilities listed in the OWASP Top Ten Vulnerabilities list, PCI DSS vulnerabilities and a range of other sensible criteria, including predictable CAPTCHA protections and open directory listings.
Wizard setup
wizard-4625603
Setting up the initial test was a very simple task. Enter a few relevant details into  ImmuniWeb’s Wizard-driven website, pay the fee and the work starts. A couple of days later a report is made available and you have around three months to download it before it is deleted automatically. You will receive warnings about the impending deletion.
The report is detailed. The first pages give an overview of the risk level based on how many vulnerabilities have been found, certain administration configuration issues that might exist and even an indication of other websites that might be impersonating yours.
Who is hosting?
report-4813405
The data in the reports is interesting and some of the issues brought to light could be easily solved. It does depend on how you have your web hosting organised, though. For example, if you run your own servers you can follow advice on upgrading certain services, such as Apache or SSH.
However, if your site runs on a hosting platform provided by a third-party, such as GoDaddy, 1&1, 123Reg or a thousand others then you have a choice: You could contact the company and request that they upgrade; or move to another host and hope that they do a better job with updates.
In this review we discovered that the hosting company we use for the SE Labs website was a little behind with some updates. We used the ImmuniWeb report as evidence that there was a potential problem and, to our surprise, the company responded fast and claimed to fix the issues.
While we could verify the changes ourselves (after all, we test security systems ourselves) we understand that for most businesses a second test would be warranted. We ran a second test for this review and were pleased to see that the previous issues had indeed been fixed.
How much?
This is where things could get expensive, though. An on-demand small business (SMB) test costs $1,499. If you are a start-up and want to have your site assessed then this is a reasonable business expense. Multiple verification tests add up, though. A faster ‘Express’ test is less expensive, coming in at $499. If you expect your site to change frequently then continuous assessments are available, with prices starting at $999 per month.
Total Cost of Reassurance
But while your site might not change, knowledge about security vulnerabilities does. New vulnerabilities are being discovered at a frightening rate and updates for popular web server components, such as MySQL, appear often. When testing our own website ImmuniWeb noted out of date software, which was updated accordingly.
By the time we ran the second test the same, updated software was again out of date. If the same issues happen to you, it might be worth learning how to test the versions of the services running at your web hosting company and give them a prod to update as and when necessary. Paying over $1,000 to assess something they should be taking care of seems unnecessary.
Monitoring the weak link
Losing control of your website is a situation no business wants to contemplate, whether it’s a start-up looking for funding or a massively profitable public company. Web application vulnerabilities are a significant weak point that can and should be assessed regularly. ImmuniWeb provides just such a service but because people are involved, as well as machine learning-equipped systems, there is a significant cost to the system, as well as an advantage over free website scanning sites and tools.
While, on the face of it, using ImmuniWeb’s service might appear expensive, compared to training your own team of penetration testers, or sub-contracting a company to do the work for you, it is good value for money.

Quantum Inside?

c0096943-quantum_computer_core-800x533-1387070

Is this the dawn of the quantum computer age? Jon Thompson investigates.

Scientists are creating quantum computers capable of cracking the most fiendish encryption in the blink of an eye. Potentially hostile foreign powers are building a secure quantum internet that automatically defeats all eavesdropping attempts.

Single computers far exceeding the power of a hundred supercomputers are within humanity’s grasp. 

Are these stories true, as headlines regularly claim? The answer is increasingly yes, and it’s to China we must look for much current progress.

The Quantum Internet
Let’s begin with the uncrackable “quantum internet”. Sending messages using the properties of the subatomic world has been possible for years; it’s considered the “gold standard” of secure communications. Chinese scientists recently set a new distance record for sending information using quantum techniques when they transmitted data 1,200Km to a special satellite. What’s more, China is implementing a quantum networking infrastructure.

QuantumCTek recently announced it is to deploy a network for government and military employees in the Chinese city of Jinan, secured using quantum key distribution. Users will send messages encrypted by traditional means, with a second “quantum” channel distributing the associated decryption keys. Reading the keys destroys the delicate state of the photons that carry them, so it can only be done once by the recipient, otherwise the message cannot be decrypted and the presence of an eavesdropper is instantly apparent.

The geopolitical implications of networks no foreign power can secretly tap are potentially immense. What’s scarier is quantum computers cracking current encryption in seconds. What’s the truth here?

Encryption Under threat
Popular asymmetric encryption schemes, such as RSA, elliptic curve and SSL, are under threat from quantum computing. In fact, after mandating elliptic curve encryption for many years, the NSA recently declared it potentially obsolete due to the coming quantum computing revolution.

Asymmetric encryption algorithms use prime factors of massive numbers as the basis for their security. It takes a supercomputer far too long to find the right factors to be useful, but it’s thought to be easy for a quantum algorithm called Shor’s Algorithm.

For today’s strong symmetric encryption, such as AES and Blowfish, which use the same key to encrypt and decrypt, the news is currently a little better. It’s thought that initially, quantum computers will have a harder time cracking these, only really halving the time required by conventional hardware. So, if you’re using AES with a 256-bit key, in future it’ll be as secure as a 128-bit key.

A Quantum Leap

2000q2bsystems2bin2blab2bfor2bwebsite-9704561

How far are we from quantum computers making the leap from flaky lab experiments to full production? The answer depends on the problem you want to solve, because not all quantum computers are the same. In fact, according to IBM, they fall into three classes.

The least powerful are quantum annealers. These are available now in the form of machines from Canada’s D-Wave. They have roughly the same power as a traditional computer but are especially good at solving optimisation problems in exquisite detail.  Airbus is already using this ability to increase the efficiency of wing aerodynamics.

More powerful are analogue quantum computers. These are much more difficult to build, and IBM thinks they’re about five years away. They will be the first class of quantum computers to exceed the power of conventional machines. Again, they won’t run programs as we think of them, but instead will simulate incredibly complex interactions, such as those found in life sciences, chemistry and materials science.

The most powerful machines to come are universal quantum computers, which is what most people think of when discussing quantum computers. These could be a decade or more away, but they’re coming, and will be exponentially more powerful than today’s fastest supercomputers. They will run programs as we understand them, including Shor’s Algorithm, and will be capable of cracking encryption with ease. While they’re being developed, so are the programs they’ll run. The current list stands at about 50 specialised but immensely powerful algorithms. Luckily, there are extremely complex engineering problems to overcome before this class of hardware becomes a reality.

Meanwhile, quantum computer announcements are coming thick and fast.

IBM has announced the existence of a very simple device it claims is the first step on the path to a universal quantum computer. Called IBM Q, there’s a web portal for anyone to access and program it, though learning how and what you can do with such a device could take years.

Google is pursuing the quantum annealing approach. The company says it plans to demonstrate a reliable quantum chip before the end of 2017, and in doing so will assert something called “quantum supremacy“, meaning that it can reliably complete specialised tasks faster than a conventional computer. Microsoft is also in on the action. Its approach is called StationQ, and the company been quietly researching quantum technologies for over a decade.

Our Universal Future

types-quantum-computers-7915887

While there’s still a long way to go, the presence of industry giants means there’s no doubt that quantum computers are entering the mainstream, but it’ll probably be the fruits of their computational power that we see first in everyday life rather than the hardware itself. So, solutions to currently difficult problems and improvements in the efficiency of everything from data transmission to batteries for electric cars could start appearing.

Life will really change when universal quantum computers finally become a reality. Be in no doubt that conventional encryption will one day be a thing of the past. Luckily, researchers are already working on so-called post-quantum encryption algorithms that these machines will find difficult to crack.

As well as understandable fears over privacy, and even the rise of quantum artificial intelligence, the future also holds miracles in medicine and other areas that are currently far from humanity’s grasp. The tasks to which we put these strange machines remains entirely our own choice. Let’s hope we choose wisely.

What is Machine Learning?

machine_learning-1991327… and how do we know it works?

What’s the difference between artificial intelligence and machine learning? Put simply, artificial intelligence is the area of study dedicated to making machines solve problems that humans find easy but digital computers find hard, such as driving cars, playing chess or recognising sarcasm. Machine learning is a subset of AI dedicated to developing techniques for making machines learn to solve these and other “human” problems without the insanely complex task of explicitly programming them.

A machine is said to learn if, with increasing experience, it gets better at solving a problem. Let’s take identifying malware as an example. This is known as a classification problem. Let’s also call into existence a theoretical machine learning program called Mavis. Consistent malware classification is difficult for Mavis because it is deliberately evasive and subtle.

silicon2bbrain-1881268For it to successfully classify malware, we need to show Mavis a huge number of files that are known to be malicious. Once Mavis has digested several million examples, it should be an expert in what makes a file “smell” like malware.

The spectrum of ways in which Mavis might be programmed to learn this task is very wide indeed, and filled with head-spinning concepts and algorithms. Suitable approaches all have advantages and disadvantages. All that counts, however, it’s whether Mavis can spot and stop previously unknown malware even when the “smell” is very faint or deliberately disguised to confuse it into an unfortunate misclassification.

A major problem for developers lies in proving that their implementation of Mavis intelligently detects unknown malware. How much training is enough? What happens when their Mavis encounters a completely new threat that smells clean? Do we need a second, signature-based system until we’re 100% certain it’s getting it right every time? Some vendors prefer a layered approach, while others go all in with their version of Mavis.

Every next generation security product vendor using machine learning says their approach is the best, which is entirely understandable. Like traditional AV products, however, the proof is in the testing. To gain trust in their AI-based products, vendors need to hand them over to independent labs for a thorough, painstaking work out. It’s the best way for the public, private enterprises, and governments to be sure that Mavis in her many guises will protect them without faltering.

About

SE Labs Ltd is a private, independently-owned and run testing company that assesses security products and services. The main laboratory is located in Wimbledon, South London. It has excellent local and international travel connections. The lab is open for prearranged client visits.

Contact

SE Labs Ltd
Hill Place House
55A High Street
Wimbledon
SW19 5BA

020 3875 5000

info@selabs.uk

Press