VIDRAFT Demonstrates Quantum Cryptanalysis of Symmetric-Key Cipher Structures on a Real 156-Qubit IBM Processor
TL;DR: Korean Pre-AGI AI startup VIDRAFT has successfully demonstrated Simon's algorithm-based quantum cryptanalysis of reduced Even-Mansour and Feistel cipher structures on IBM's real 156-qubit
ibm_kingstonquantum computer — not a simulator. This is a proof-of-concept (not a break of production ciphers like AES), but it represents a meaningful milestone for running attack algorithms at unprecedented scale on noisy real hardware. VIDRAFT has also open-sourced a browser-accessible quantum cryptanalysis toolkit on Hugging Face covering five canonical cipher structures.
What it is
VIDRAFT — a Korean Pre-AGI AI startup known for its Darwin language model family and the AETHER next-generation architecture — has conducted what it describes as the first public demonstration of quantum cryptanalysis against symmetric-key cipher structures on real quantum hardware at this scale.
The experiment targeted two foundational building blocks of modern symmetric cryptography:
- Even-Mansour construction — a minimal, keyed permutation structure that underlies many block ciphers
- Reduced Feistel construction — a 3-round skeleton of the Feistel network, the structural basis of ciphers like DES (though importantly, this is not full DES)
The hardware used was IBM's 156-qubit ibm_kingston quantum processor — real quantum hardware, not a simulator.
VIDRAFT was explicit about scope: this is not a break of AES, bank encryption systems, or any production cipher. It is a structured proof-of-concept targeting reduced/skeletal cipher constructions.
How it works
The cryptanalytic approach is based on Simon's algorithm, a quantum algorithm designed to find hidden periods in functions — a property that maps directly onto the structural weaknesses of certain symmetric cipher constructions.
At a high level, the experiment works as follows:
Hidden period recovery: Simon's algorithm is applied to find the secret-key-determining hidden period embedded inside the cipher structure. For Even-Mansour, this was demonstrated across block sizes N=5 through N=10. For the 3-round reduced Feistel structure, hidden periods were recovered under block sizes of 6 and 8 bits.
Self-verification procedure: Critically, VIDRAFT applied a self-verification step where, for each experimental instance, a second independent key was recovered in parallel. This is designed to demonstrate that the result is not reverse-engineered from a known answer — the hidden structure is genuinely being discovered from raw quantum execution output.
Noise mitigation (not full QEC): The experiments ran without full quantum error correction (QEC). Instead, noise mitigation techniques were applied to manage the inherent noise of NISQ (Noisy Intermediate-Scale Quantum) hardware. VIDRAFT frames this not as "quantum beating classical" but as demonstrating that attack algorithms can be executed at previously unseen scale on noisy real hardware.
The key conceptual claim: the hidden periodicity that underlies these cipher constructions — the mathematical structure that Simon's algorithm is designed to exploit — was extracted from actual quantum hardware execution, not classical simulation.
Benchmarks & results
Based on the publicly reported results:
- Even-Mansour: Hidden periods (determining the secret key structure) successfully recovered for block sizes N=5 through N=10
- Feistel 3-round reduced structure: Hidden periods confirmed at block sizes 6 and 8
- Self-verification: Independent second-key recovery applied per instance to validate authenticity of results
-
Hardware: IBM 156-qubit
ibm_kingston(real quantum processor) - Error handling: Noise mitigation only — no quantum error correction
No classical baseline comparison figures, qubit circuit depth counts, or shot/fidelity statistics were disclosed in the source article. VIDRAFT has stated it plans to pursue formal paper publication and external validation of these results.
How to try it
VIDRAFT has made a quantum cryptanalysis toolkit publicly available on Hugging Face, described as a "QuantumOS Simulator." It covers five representative cipher structures:
- Linear cipher
- Block cipher / SPN structure
- Even-Mansour
- CBC-MAC
- Feistel structure
The toolkit is browser-accessible — engineers can interactively observe how each cipher structure can be made vulnerable by quantum algorithms without any local setup required.
To find it, search for VIDRAFT on Hugging Face. The specific model/space slug was not disclosed in the source article, so check VIDRAFT's official Hugging Face organization page directly.
No GitHub repository URL, API endpoint, or installable package was mentioned in the source coverage at time of writing.
FAQ
Q: Does this mean AES or real-world bank encryption is broken?
A: No. VIDRAFT explicitly stated this does not break AES or any production banking cipher. The targets are reduced structural skeletons — 3-round Feistel and Even-Mansour with small block sizes — used as proof-of-concept vehicles. Full production ciphers operate at far larger scales and require fault-tolerant quantum hardware that does not yet exist.
Q: Why does Simon's algorithm matter for symmetric cryptography if Grover's algorithm is the usual reference?
A: Grover's algorithm provides a quadratic speedup for brute-force search and is the common reference for symmetric cipher threat modeling. Simon's algorithm is exponentially faster for a different problem: finding hidden periods in functions. Certain symmetric constructions — particularly Even-Mansour and Feistel-based designs — expose structural period-finding vulnerabilities that Simon's algorithm can exploit directly, making it a more targeted (and theoretically more powerful) tool for those specific structures.
Q: Is this relevant to post-quantum cryptography (PQC) standardization work?
A: Indirectly yes. The PQC standardization effort (NIST, etc.) has focused heavily on asymmetric cryptography threats from Shor's algorithm. This research highlights that symmetric cipher structures also carry quantum-exploitable structural properties worth studying — particularly as quantum hardware scales up toward fault tolerance.
Originally reported by 전자신문 (2026-07-06) — source article.
Top comments (0)