A long-standing concern among cybersecurity experts about quantum computing is that these systems will eventually achieve enough processing power to break classical RSA encryption. This possibility was famously revealed by Shor’s algorithm 30 years ago, but it still looms as an overlooked risk that today’s quantum computers are not only potential platforms for attack, but also vulnerable targets.
The two researchers believe that while the focus on the need for strong post-quantum cryptography (PQC) is an important issue, it should not overshadow the risks that quantum computing systems themselves face from cyberattacks. At the Black Hat USA 2024 conference in Las Vegas next month, Adrian Colesa, senior security researcher at Bitdefender, and Sorin Bolos, co-founder and software engineer at Transilvania Quantum, will discuss the risks of quantum vulnerabilities and their real-world implications.
Risk assessment of post-quantum computing platforms
Boros and Kolesa will present the white paper’s findings on Thursday, August 8, during a session titled “From Weapons to Targets: The Quantum Computing Paradox.”
“When most people think about quantum computers and security, they think about Shor’s algorithm and the fact that if you have a powerful enough quantum computer, you can use Shor’s algorithm to factor and break encryption,” Boros says. “But we’ve flipped that around and said, ‘What about the quantum computer itself? How secure is it? Can you attack it?'”
As a Romanian startup that developed Uranium, an open-source quantum computing platform for prototyping quantum algorithms, Boros decided to hire Transilvania Quantum to investigate the security risks of its quantum computing infrastructure. “We only had quantum expertise, but no cybersecurity expertise, so we turned to Bitdefender,” he says.
Last October, the two researchers began leveraging their complementary expertise in cybersecurity and quantum computing, with Transylvania focusing specifically on attacking quantum computers from IBM and IonQ, as well as quantum software development kits such as Qiskit.
As a provider of endpoint protection, cloud and managed cybersecurity tools, Bitdefender had quantum expertise around Transilvania’s PQC focus.
“The Bitdefender team looked at traditional attack vectors, such as attacking end-user systems or quantum development software that could be subverted by an attacker, and then looked at how cloud services that provide access to quantum computers could be attacked,” Kolesa explained.
Discovering weaknesses in quantum bits
Boros said they investigated imperfections in quantum bits, or qubits, the quantum computing equivalent of bits in classical computing. Their work explored the possibility of unwanted interactions, vulnerabilities to prompt injection, and other attack surfaces that are prevalent in classical computing environments.
“We adapted our attack for the quantum world and ran experiments,” Boros said.
Boros said organizations using quantum computing capabilities currently access them through quantum service providers, which are integrated platforms hosted on cloud services such as Microsoft Azure or Amazon Web Services, or by companies hosting their own quantum clouds.
In recent years, well-funded organizations have begun researching how quantum computing can help handle complex computational workloads that are beyond the capabilities of the most powerful classical systems.
These include companies involved in drug discovery and medical research, such as Amgen, Cleveland Clinic, Merck, and Johnson & Johnson. Additionally, most of the world’s largest financial services providers, including Bank of America, JP Morgan Chase, and Wells Fargo, have launched research initiatives aimed at creating financial models that cannot be achieved with traditional computing techniques. All of this could be attractive targets for cybercriminals.
But the two researchers note that security often takes a backseat as these organisations aim to beat competitors with new breakthroughs in drug discovery, financial models and other areas.
Kolesa said he broke down his research into four ways that attackers could target quantum computers.
An attack on a quantum computer launched from a classical system.
An attack that manipulates the quantum processing unit (QPU) of qubits.
Attacking the QPU with quantum components;
Attacks on RSA encrypted data.
Many of the vulnerabilities discovered in quantum computing systems share the same characteristics as classical computing environments, meaning similar countermeasures are required.
“For example, checking whether a software development kit (SDK) comes from a trusted source, or whether the transpiled source comes from a trusted source. [the quantum equivalent of compiled] “This circuit is exactly what should be sent to a quantum computer,” Kolesa says.
As quantum computer capacity continues to grow beyond 1,000 qubits, providers will need to focus on error correction – the process of identifying the root cause of risk to an organization, Boros warned.
“Errors can be inserted by someone or arise naturally from the environment,” he says. “Error correction is one of the key aspects of protecting against malicious users.”
