A team of physicists at ETH Zurich says it has demonstrated a way to generate certifiably perfect random numbers using quantum mechanics, a development that could strengthen the foundations of future quantum-secure encryption systems.
The research, published in the journal Nature, addresses a long-standing problem in cybersecurity: truly random numbers are difficult to create in the real world, yet modern encryption systems depend on them. Even tiny statistical biases in random number generators can create openings for attackers, particularly in high-security environments such as financial systems, military communications, digital identity platforms and quantum-safe networks.
The ETH Zurich researchers said in a university news release their method can take imperfect randomness and amplify it into fully random sequences that can be mathematically certified.
The work comes as governments and technology companies prepare for the arrival of quantum computers capable of breaking parts of today’s public-key cryptography. Much of the discussion around post-quantum cybersecurity has focused on new encryption algorithms resistant to quantum attacks. But encryption systems also rely on another ingredient: unpredictable random keys.
If the randomness behind those keys is flawed, the security of the entire system can weaken regardless of how advanced the encryption algorithm may be.
The ETH Zurich team used a quantum phenomenon known as entanglement to address the problem. Their experiment relied on two superconducting quantum chips connected by a 30-meter cryogenic link cooled to temperatures near absolute zero.
Each chip contained a quantum bit, or qubit, capable of existing in multiple states simultaneously. Microwave photons traveling between the chips created entanglement, linking the qubits so that measurements performed on one affected the outcome of the other.
The separation between the qubits was designed to ensure that no information could travel between them during the measurements, even at the speed of light. According to the researchers, this helped preserve the integrity of the randomness.
The experiment also incorporated a Bell test, a method used in quantum physics to confirm that correlations between particles cannot be explained through classical physics alone. The researchers said improvements in both data quality and measurement speed allowed them to achieve randomness amplification at a level not previously demonstrated.
According to the research team, the measurement settings themselves were selected using an imperfect random number generator. The resulting outputs were then processed through a specialized algorithm that amplified the weak randomness into statistically perfect random numbers.
Renato Renner, a professor in ETH Zurich’s Department of Physics, said the result effectively creates random sequences that remain unpredictable regardless of future analytical methods.
The researchers compared the potential role of such systems to atomic clocks, which provide highly precise timing standards for digital infrastructure. In this case, the quantum system could serve as a physically certified source of randomness for cybersecurity and other digital applications.
Possible uses include encryption services, secure communications, blockchain systems, lotteries and distributed public randomness services. The approach could also become important for quantum networks and future communications infrastructure designed to withstand attacks from quantum computers.
The work also highlights the growing overlap between quantum computing research and cybersecurity infrastructure. While quantum technologies are often discussed as future threats to encryption, the same physics is increasingly being explored as a tool for building stronger security systems.
The researchers acknowledged that the setup remains highly specialized and requires superconducting hardware and cryogenic cooling systems. Still, the experiment demonstrates that certified randomness generation can move beyond theory and into practical implementation, an advance that could influence the design of future quantum-secure systems.
Image: A sheep image encrypted using ordinary randomness (center) and certified perfect randomness from the ETH experiment (right). Only perfect randomness turns the image entirely into noise. (ETH Zurich)
