Was a wire transfer instruction or a request to access critical infrastructure really sent from a secure terminal inside the company, or by an attacker on the other side of the world? Conventional digital authentication can confirm possession of a password or cryptographic key, but it struggles to guarantee the physical location where an operation was carried out. An international research team including Krister Shalm and Gautam A. Kavuri of the U.S. National Institute of Standards and Technology (NIST) has now demonstrated, in the laboratory, "quantum position verification" (QPV), which combines the properties of quantum mechanics with the speed-of-light limit from relativity to address this problem. In a paper published in Nature on November 18, 2026 (DOI: 10.1038/s41586-026-11247-3), the team reports the first demonstration of a complete position-verification protocol in a "device-independent" framework, one that does not trust the internal workings of the remote quantum devices.

AD

How quantum position verification removes the need to trust the device

Quantum position verification (QPV) is a technique in which two trusted "verifiers" work together to check whether a remote "prover" has performed a requested quantum operation within a specific target region. Earlier approaches to QPV had to assume that the prover's quantum hardware had not been tampered with, or had to place limits on an attacker's computing power or quantum computer capabilities. Previous experiments implemented only parts of a protocol and had not achieved a complete demonstration that does not trust the devices.

This study treats the prover's quantum devices as black boxes and verifies position without relying on their internal structure or operating model, an approach known as "device-independent." The classical equipment the verifiers use for timing and recording is still assumed to be trustworthy. Security also requires a limit on the amount of entanglement attackers can share before an attempt begins. Position verification under these conditions is made possible by a "loophole-free Bell test," which confirms correlations produced by quantum entanglement, together with the physical law that no signal can travel faster than light.

What is verified is that, within a specific region of spacetime, the quantum operations required by a position-dependent protocol were performed. It does not identify who performed the operation, where that person is continuously located, or where any other devices they own are. It is a different technology from continuous GPS location tracking or identity verification using biometric information.

Verifying position with photons and random numbers across a 195-meter network

The team set up two verification stations 195.1 ± 0.3 meters apart on the NIST campus in Boulder, Colorado. The stations were connected by electrical cables and thin optical fibers running through a corridor of a laboratory building, and the target region where the prover was supposed to be located was set midway between them. The experiment was conducted in a controlled laboratory environment, not on a city-scale commercial quantum network.

In the verification procedure, one verifier first generates pairs of photons in an entangled state, keeps one photon of each pair and sends the other to the prover. The measurement results of entangled particles show strong correlations that classical physics cannot explain, no matter how far apart the particles are. However, these correlations cannot be used to communicate information faster than light.

Next, each of the two verifiers sends an unpredictable random number (a classical challenge bit). The transmission times are precisely adjusted so that both numbers arrive at the claimed target region at the same moment. The prover uses the two numbers to compute which basis to use when measuring the polarization (the direction in which the light's electric field oscillates) of the photon it received. It measures the photon according to that setting and sends the result back to both verifiers.

The verifiers check how long it took from sending the random numbers to receiving the measurement result. Because signals cannot travel faster than light, the response time narrows down the region where the operation could have taken place. The verifiers also examine the statistical correlation between measurement settings and results for the photons they kept and those measured by the prover, confirming a violation of Bell inequalities without loopholes. This correlation cannot be reproduced by classical mechanisms, and quantum information can neither be freely copied nor have the same quantum correlations shared without limit among multiple attackers. Combining these properties with the speed-of-light constraint allows the scheme to detect attempts to spoof a position from outside the target region.

According to the preprint the team posted on arXiv (arXiv:2601.16892), each protocol instance consisted of 30 million trials over 2 minutes for the basic protocol, and 60 million trials over 4 minutes for the extended protocol, which also covers attackers who share entanglement in advance.

AD

Pass rates up to 99%, and better localization than classical protocols

Over two days, the team evaluated a total of 335 runs: 232 of the basic protocol and 103 of the extended protocol. In the basic protocol, 224 of 232 runs passed, a pass rate of 96.5%. In the extended protocol, 102 of 103 runs passed, or 99.0%. The design target for completeness, meaning the probability that an honest prover passes, was set at 0.97725 (97.725%), equivalent to two standard deviations of a Gaussian distribution, and the measured results agreed well with this target. The soundness parameter, which sets an upper bound on the probability of wrongly accepting a dishonest prover, was set at . This is not a measured pass rate but a security criterion against the assumed attacker model.

Protocol / comparison Attacker / comparison conditions Measured / certified result
Basic protocol (232 runs) No pre-shared entanglement before trials; 2 minutes and 30 million trials each 224 runs passed (96.5%)
Extended protocol (103 runs) Upper bound on average robustness of pre-shared entanglement per trial of ; 4 minutes and 60 million trials each 102 runs passed (99.0%)
Comparison with 1D classical protocol Ideal classical position-verification protocol Quantum protocol confines the target region times more tightly
Comparison with classical protocol of identical delay Classical protocol with equal communication and processing delays Quantum protocol confines the 3D target region volume times more tightly

The quantum approach also showed an advantage over classical position verification. According to the paper, compared with the theoretically best ideal classical position-verification protocol in one dimension, the successful quantum protocol narrowed the region in which the operation took place by a factor of . Compared with a classical protocol with the same communication and processing delays, it reduced the three-dimensional volume of the target region by a factor of .

However, these factors do not indicate by how many meters the error was reduced, as with a GPS-style position measurement. Nor did the placement of the verifiers, 195 meters apart, change. What is being compared is the size of the target region that each protocol, classical or quantum, allows.

The scope of the demonstrated security, and its limits against stronger attacks

The security guaranteed by this experiment comes with clear conditions. Attackers are assumed to be able to use unlimited classical and quantum computing power and communication during each trial, but the amount of entanglement they can share before a trial begins is limited. The security analysis of the basic protocol targets attackers who share no entanglement in advance. The extended protocol permits a certain amount of pre-shared entanglement, but the average "robustness of entanglement" per trial across an entire instance is set at or less.

Robustness of entanglement is a measure of how strong the entanglement in a quantum state is. For comparison, the paper gives the robustness of the entanglement generated per trial by the experimental source as , and that of an ideal Bell state as $1$. The pre-shared attacker entanglement assumed in the extended protocol is very small even compared with these values.

In fact, the authors cite a method called the "garden-hose attack" and state that attackers who share two qubit pairs, each with robustness , could break the protocol as implemented. At the same time, the team proved a general theorem showing that increasing the number of classical challenge bits sent by the verifiers can counter attackers with far greater quantum resources. The security of this extended scheme has been shown theoretically, but the experiment did not implement that configuration.

The experiment also did not detect or block intrusions by real-world cyberattackers. The protocol's security was evaluated by analyzing the collected measurement data against the assumed attacker model. One key technology behind the successful demonstration is high-efficiency photon detection. If too many photons are lost, even an honest prover cannot meet the Bell-test pass conditions and cannot prove it is within the target region.

AD

A decade of photonics progress behind the demonstration, and the road to practical use

The result builds on loophole-free photonic Bell-test technology established in work including the NIST-led research of 2015. The findings were released as a preprint in January 2026 and published in Nature in November of that year. According to the official NIST announcement, the experiment depended on a fast, high-efficiency entangled-photon source developed by NIST researchers about ten years ago, and on superconducting nanowire single-photon detectors (SNSPDs) that can detect nearly every incoming photon.

NIST physicist Krister Shalm said the result shows "a way to tie digital security to the physical world." The team envisions applications such as financial transactions, legal contracts and the protection of critical infrastructure, but no demonstrations have been carried out for these uses. Scott Glancy, a NIST physicist and collaborator, estimates that real-world applications could take more than ten years.

For example, verifying that a bank customer is truly communicating with someone inside the bank's building would require a large-scale quantum communication network capable of exchanging entangled photons between the two parties with high fidelity. The demonstration confirms that, against attackers with limited pre-shared entanglement, the location of an operation can be verified without trusting the remote quantum devices. Practical use will require not only building out such quantum communication infrastructure but also implementing protocols with multi-bit challenges that can withstand attackers who share more pre-existing entanglement.