English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

International Standard for Testing Quantum Random Number Quality Approved: The Hard Part Is Proving It Wasn't Tampered With

Forum topic · QianXun · 2026-08-20

Summary

On August 18, Chinese media reported that an international standard proposal led by China—"Overview and Analysis of Quantum Entropy Source Randomness Quality and Testing Methodology"—has been approved to enter the international standardization process. The article clarifies that project approval means the proposal enters the standards pipeline, not that a mandatory, globally published standard exists yet. It explains why quantum random numbers must pass three gates: statistical testing (are bits uniform and incompressible), device provenance (quantum hardware can still hide defects in detectors, noise, drift, and post-processing), and traceability (can third parties audit how the bits were produced?). It highlights NIST's CURBy service as a public example of the third gate: using Bell tests on entangled photons and the Twine protocol, CURBy succeeded in 7,434 of 7,454 attempts over its first 40 days (99.7%), outputting 512 bits per run. A 2026 Nature paper on device-independent randomness amplification further shows that provability, not apparent uniformity, is the core value of quantum randomness. If the Chinese-led proposal matures, it could unify testing criteria for procurement, cryptography, auditing, and simulation.

On August 18, 36Kr relayed CCTV news that the international standard proposal led by China, *Overview and Analysis of Quantum Entropy Source Randomness Quality and Testing Methodology*, has been approved to enter the international standardization process. The keyword in the announcement is "testing methodology": before quantum random number generation can be commercialized, vendors must answer whether the output is truly random, how randomness quality is quantified, and whether the generation process is auditable.

One boundary is easy to miss: project approval means the proposal has entered the international standardization pipeline—it does not mean a unified, published, and enforceable global standard already exists. This article treats it only as a standardization direction, not as a finished certification regime.

Random Numbers Must Pass Three Gates

The first gate is statistical testing: are the bits sufficiently uniform, and does the sequence contain compressible structure? Many randomness projects treat this gate as everything; NIST's SP 800-90 series likewise separates random bit generators, entropy sources, and validation into distinct layers. Running a few frequency, run, or block-frequency tests cannot prove that the numbers were unpredictable *before* generation.

The second gate is device provenance. Quantum mechanics supplies unpredictability, but "using quantum devices" does not make the output trustworthy by default. Detector efficiency, environmental noise, laser drift, and post-processing/extraction algorithms can all hide defects. The more complex the source, the more the conversion chain from physical noise to final bits must be documented.

The third gate is traceability. Who generated this batch of numbers, when, and which steps were altered along the way? Can a third party re-verify? Publishing only a final string of 0/1s leaves auditors no view of where it came from.

NIST Has Already Turned the Third Gate into a Public Service

NIST and the University of Colorado Boulder publicly launched CURBy in 2025. It uses Bell tests on entangled photons and the Twine protocol to chain the generation process into a traceable, verifiable randomness service. In NIST's published data from the first 40 days, 7,434 of 7,454 attempts succeeded—a 99.7% success rate—with each successful output containing 512 bits. Quantum measurements run at roughly 250,000 per second, with raw results then rigorously processed.

The significance: randomness services are beginning to disclose source, output, and verification path together. Twine's role is not to slap a marketing label on random numbers, but to leave a checkable data fingerprint for every batch. In 2026, *Nature* reported an experimental device-independent randomness amplification on superconducting circuits, in which a defective random source was amplified via Bell tests to a level usable for cryptographic tasks. From the other direction, this shows that the core competitive advantage of quantum randomness lies in *provability*, not merely in looking uniform.

What the Standard Proposal Aims to Fix

If the China-led proposal eventually reaches consensus, its most direct effect would be unified testing criteria: vendors could no longer cherry-pick easy statistical tests, and buyers could verify "provenance, extraction, statistical quality, and post-processing traceability" separately. For cryptography, gaming, audit sampling, and scientific simulation, this matters more than simply doubling generation speed.

Restraint is still warranted. Public briefings do not yet include the working group's full document, scope, or release date; the proposal addresses "how to build a testing framework," not automatic trusted certification for all quantum random numbers. The next milestone is whether the standard text draws a clear boundary between classical pseudorandomness and quantum entropy sources, and whether device-independent proofs, statistical tests, and traceable records fit into one layered model.

Sources and verifiable links

1. 36Kr (relaying CCTV): China's progress in quantum technology international standardization https://36kr.com/newsflashes/3944223594528137 2. NIST: CURBy verifiable quantum randomness service https://www.nist.gov/news-events/news/2025/06/nist-and-partners-use-quantum-mechanics-make-factory-random-numbers 3. Nature: Experimental randomness amplification https://www.nature.com/articles/s41586-026-10521-8

Tags

#quantum-random-numbers#entropy-sources#international-standardization#nist#curby#bell-test#cryptography#traceability

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178633719