Q-Day is coming and we are not prepared
We have a problem.
Our advances in information technology have transformed our world and what’s possible. Forget those futuristic “videophones” that Jack Bauer made look so cool talking to his CTU colleagues in episodes of 24, I can now Skype with my parents in full HD video on my smartphone sitting at the beach. Forget the video gaming LAN parties of my youth, kids can now play multiplayer video games with almost anyone around the globe. And forget waiting in line at the bank and writing cheques, I can easily make payments, check balances, shop, and invest all online with a couple of clicks.
But that all sounds great, what’s the problem? It’s simple, the security of our new information technology has not kept pace.
We’re not as secure as we think.
Individual hacking attacks are so numerous, from the Sony hack of 2011 to the Talk Talk and Ashley Madison hacks of 2015, that we’ve almost become desensitised to them. And while those certainly weren’t good, it’s the new generation of hacking that should have us really worried. In 2015, security engineer “white hat” hackers sitting comfortably in their living room, remotely shutdown a Jeep Cherokee with a Wired journalist sitting in the driver’s seat, while he was driving along at 60 mph on a St. Louis highway! In 2016, during the middle of winter, a cyber attack took out 1/5th of Kiev’s power grid. What’s worse… this wasn’t the first time it happened! More recently, researchers have discovered a systematic series of attacks against power plants, refineries, and other critical infrastructure. Here hackers have been targeting safety systems setup to prevent health and life-threatening accidents! Recently, these same hackers have escalated their activities and begun probing US power grids. With all of this, it’s no wonder the US recently declared a national emergency over IT threats and expressed concerns over potential state actor backed hacking.
New capabilities, new threats.
Hackers are currently selling a variety of hacking and malware tools on the dark web starting from the bargain basement price of $1. This is big business, with some hackers raking in $80,000 a month or more causing more than 1 billion people to have their data compromised in 2018 alone. And while this should already disturb you, it’s the rise of state sponsored hacking with new tools that’s the real problem. For example, researchers are now applying new techniques like machine learning to the cryptanalysis of random number generators, both classical and quantum, exploiting patterns that many of the traditional suite of tests can miss. But the real kicker is something called quantum computing.
Quantum computing is the next advancement in computing, where quantum systems are now used to represent information not in bits anymore but now in quantum bits – qubits for short. These quantum systems operate according to the laws of quantum mechanics. Typically very small scale systems or particles, good examples are single photons, the spins of electrons or artificial atoms, or trapped ions – all of which people are trying to build quantum computers with. The advantage of using qubits that operate according to new quantum computing rules is they can be used to process information in new ways, potentially much more efficiently.
This new computing paradigm promises great advancements, such as quantum simulations that could allow the efficient design of new molecules, fertilisers, and cancer drugs. Quantum computing also happens to be tailor made to solve many of the mathematical problems which currently underpin the security of our current, convenient cryptographic algorithms. The RSA algorithm and elliptic curve cryptography both depend on the difficulty of a particular mathematical problem, called the hidden subgroup problem, for their security. But in 1994, Dr. Peter Shor, in what’s now famously known as Shor’s algorithm, showed an algorithm that could be run on a quantum computer that could efficiently solve exactly this problem! Thus, while the development of a quantum computer will have all sorts of positive effects it will also radically speed-up the cracking of our current encryption codes.
Quantum computing is closer than we think.
Quantum computers might seem futurist and far enough away that we don’t have to worry, but that would be a mistake. A little thought shows just how dangerous this is and that we should be starting to worry a lot sooner – indeed, D day might already have passed!
A nifty little equation, termed Mosca’s equation, sums up nicely when we need to worry about upgrading our cyber security. It’s given by
x + y > z
where x = the security lifetime of our data (i.e. how long we want it to be secure), y = the upgrade time (i.e. the time needed to transition our information technology security systems to quantum-safe ones), and z = the time to build a quantum computer (i.e. the time at which our current security systems will become insecure). In a nutshell? If it’s going to take 10 years to upgrade our security systems and you want your medical records to be secure for 10 years at least, meanwhile there’s a reasonable chance a quantum computer is going to be built in the next 15 years, then you’re already out of luck. Since there’ll be 5 years while your sensitive data is effectively unencrypted and in the clear. See here for a whimsical article explaining this “store now, crack later” attack using the recent Avengers Endgame movie.
While most of the quantum computers we’ve build so far have been small prototypes, quantum computing is what’s known as an exponential technology; that is, progress isn’t linear. The human genome project was another great example of an exponential technology. It was a multi-billion dollar project which spanned 13 years. Famously, it had very little to show for itself more than half-way through the project. It certainly didn’t have half the human genome mapped. But all of a sudden the investments and work started to pay off. The technologies in development started to produce results that snowballed one on top of the other, which fed back to produce faster and faster development. And in a few short years, the human genome project completed with a resounding success. What’s more the cost has also dropped astronomically, from hundreds of millions of dollars at the turn of the century to now less than $1,000 in 2019.
For quantum computing, we’re now starting to see the same behaviour after many years of seemingly small-scale development in university labs. We’re starting to see the exponential “kink” in the graph. Where 2, 4, and then 8 qubits was the record for a very long time, in the matter of a few short years we’ve seen 19 qubits (Rigetti, 2018), 49 qubits (Intel, 2018), 50 qubits (IBM, 2017), and 72 qubits (Google, 2018) now demonstrated. Even better, quantum computers are starting to actually be useful in commercial applications, such as BT using a DWave machine to optimise it’s network planning and Accenture partnering with 1Qbit and Biogen to develop a quantum-enabled molecular comparison application. And it’s this last point that will really see the technology take off, once businesses start to see a commercial gain from using the technology, investment and development will simply re-enforce one another and take off.
Don’t believe me? Try plotting the trajectory of commercial system development in terms of number of qubits. Far from the shallow trajectory traced by academic systems, you’ll see that commercial quantum computers (albeit in the very early days with a few data points) are very nearly following their own Moore’s law (see upcoming KETS’ Threat Assessment). The potential is so strong that some experts think there’s a 1 in 6 chance that by 2026 a quantum computer will be built that is able to break RSA-2048.
Why we should be worried?
So just how worried should we be? Very, as several recent analyses show. Researchers are working the problem from both ends. As we’ve just seen, the development of quantum computers is speeding up exponentially. Similarly, researchers are making great strides in optimising the algorithms. Initial estimates in 2015 of how many qubits it would take to factor a 2048 bit number on a quantum computer were about 1 billion. But only a few short years later, researchers have brought it down to about 20 million qubits to factor a 2048 bit number in about 8 hours. Factoring is roughly equivalent to the hidden subgroup problem and is used by the RSA algorithm for its security. Currently, industry is still typically using key sizes of 1024 bits, though many are now starting to switch to 2048 bit keys based on NIST’s recommendations.
Recent work by Prof. Michele Mosca (whose equation we met above), a computer scientist, quantum information specialist, and Deputy Director of the Institute for Quantum Computing and collaborator Dr. Vlad Gheorghiu also analysed the time and resources needed for quantum cryptanalysis of some of our current cryptographic schemes. What they found was that with some very modest assumptions about quantum computing hardware, RSA-3072 and the NISTS-256 elliptic curve algorithm (roughly the equivalent to RSA-3072) should be crackable by 2042 within less than 1 day. And if one allowed the quantum computer to work for a year – a modest amount of time for determined state actors trying to access high value data – even larger key sizes of RSA-4096, RSA-7680, and NISTS-521 are well within reach
Now put that into the context of how long you’d like to keep your data safe for. Starting with the Top Secret information of governments. The US and UK expect classified information to remain secure for a minimum of 25 years. Thus, information sent now is expected to stay out of the public domain until at least 2044. But if the analyses above are correct, then we are already too late then in securing today’s most confidential information!
IP and commercially sensitive data (such as oil deposits, drug trial data, merger plans, and trade secrets) are expected to be protected for 5-20 years. Satellite systems, which are receiving a lot of attention at the moment as the commercial space race takes off, take 4-5 years to build and launch and are then expected to have a service lifetime of roughly 15 years. Third-party information (such as bank details, transaction data, and credit cards) is expected to be secure for 7 years or longer. And now, thanks to GDPR, disclosure carries severe consequences – with potential fines of up to 4% of a firm’s annual turnover or €20m. While some of these might still have a short security window, it’s clear that the time is coming rapidly where we can’t meet these requirements with our current solutions.
The future promises big advancements… with huge accompanying risks if we’re not careful.
So far we’ve just been talking about current technology, but the future promises advancements that will transform our lives, society, and life as we know it. Imagine when we can finally put our medical records fully online, so that if I’m in an accident on vacation and need medical care the doctors can have up-to-date access to my medical records and allergies before they treat me. Or soon we might each be mapping our own personal genome in the hopes that it can help cure a future cancer.
Major banks are moving the data and services for their entire organisation into the cloud, while new distributed ledger technology (aka Blockchain) is rapidly changing their verification services and sometimes removing the need for them completely. Similarly, some functions typically run by governments are now being stored as transactions on a public ledger that lives in cyber space for anyone to inspect. Edge computing – putting data and processing much closer to the user or application at the edge of the network – is getting us better Netflix binging and the potential for completely new services. My next car might no longer come with a steering wheel as standard. And we’re stepping into our AR and VR future, which lets us fend off a shark while cage diving or walk on Mars with friends in our living room on a Friday night.
These are fantastic advances, but my medical records and certainly my genome are… well me! How secure is my bank account living in the cloud? Bitcoin, despite its increasingly widespread adoption, is built on cryptographic primitives whose security is rapidly being undermined. Edge computing is just that, a lot of precious data sat on the edge of a network outside the traditional secure boundaries of an organisation. What happens when the first autonomous car gets sick with a virus? And how do you police a virtual environment?
Your next Xbox or PlayStation gamer tag might require more vetting then your driver’s licence.
I absolutely want my medical records to be secure. The tiny size of my bank account, currently its best protection, won’t stand up much longer when attacks can be efficiently launched at scale. All that great blockchain technology, is based on fundamentally vulnerable cryptography. Data is the new gold and absolutely must be protected out in the wild. Future cars should not only get me from A to B safely, but they should not be able to be used remotely as weapons in future terrorist attacks. And the realism that AR/VR is reaching, its biggest strength, absolutely shouldn’t allow a sexual assault to be possible from anywhere on the planet on Xbox’s and PlayStation’s networks.
Next generation solutions.
Despite what some will tell you, the solutions to all of these new problems are complex…. because our information technology is complex. And while it will likely be the large companies and organisations that feel the pain first, we as individuals won’t be far behind. We already have huge problems in front of us with how social media systems such as Facebook and Instagram should evolve in a world where state actors are using them to influence elections.
Security, long an afterthought and an annoying cost, now needs to come to the fore and be embedded in every piece of new technology as one of the first thoughts, not the last. And no one technology holds the answer, just like today, complete solutions will be made up of many different pieces: new trust structures and algorithms, hardware and software, classical and quantum. At KETS, we’re developing some of these new tools, quantum encryption tools, to build the solutions of the future.
Chris Erven, June 2019