QSignAI: Quantum-Randomness-Seeded Identity Signatures at the Intersection of AI for Science and Science for AI
Source: arXiv:2605.27729 · Published 2026-05-26 · By Dongping Liu, Aoyu Zhang, Luyao Zhang
TL;DR
QSignAI addresses the integration of quantum randomness and artificial intelligence (AI) in a deployed social participation system, leveraging recent breakthroughs recognized by the 2024–2025 Nobel and Turing awards. The work tackles three core questions: embedding real quantum randomness generation into an AI-driven high-scale social platform with acceptable latency and cost; enabling an AI bot to make quantum phenomena accessible and perceptually meaningful to general audiences; and demonstrating the practical viability of a combined system. The platform uses a Telegram bot routing participant messages through two quantum circuits running on the AWS Braket state-vector simulator to generate unique, quantum-randomness-seeded identity signatures, visually encoded as color badges linked to quantum entanglement metrics. Deployment experience shows the quantum components integrate asynchronously with minimal user-perceived latency, and the combination creates a property neither quantum randomness nor AI alone achieve. While rigorous quantitative comparisons are deferred to future work, the system offers a first-of-its-kind production example of bidirectional AI–quantum cultural and technical interplay.
Key findings
- Quantum circuit execution latency on AWS Braket SV1 simulator ranges from 2 to 8 seconds per circuit, with asynchronous integration avoiding blocking user experience.
- Circuit A uses a 4-qubit quantum random number generator producing a quantum number qnum ∈ [0, 1000] from 100 measurement shots.
- Circuit B generates a 2-qubit Bell state |Φ+⟩ with 200 measurement shots, providing an entanglement probability vector β for entropy and visual encoding.
- Quantum-randomness-seeded ToyLWE identity signatures are derived with SHAKE-256 hashing mixing username seed, qnum, and a fresh random nonce.
- Visual identity badges encode qnum and Bell state probabilities as Hue-Saturation-Lightness (HSL) colors with distinct perceptual spacing using the golden angle (137.5°).
- The Telegram bot platform supports 1B+ monthly active users, uses a multi-layer infrastructure on AWS, and maintains security best practices including secret validation and encrypted storage.
- Fallback to local SHAKE-256 pseudo-random signatures is triggered after 30s quantum task timeout, ensuring uninterrupted UX and auditability of provenance ('local-fallback' flags).
- Deployment demonstrates feasibility of a bidirectional AI-quantum system combining quantum-derived identity with conversational AI mediation and public visual quantum literacy cues.
Threat model
The adversary is considered classical and unable to replicate or predict true quantum measurement outputs, consistent with quantum information theory. They cannot reproduce the quantum-randomness-seeded identity tokens since the randomness is irreproducible by classical means. Attackers cannot access secret quantum measurement results or internal states to reverse-engineer the identity signatures. However, no advanced attacker models or side-channel attacks on the bot messaging or infrastructure are assumed or evaluated in depth.
Methodology — deep read
The threat model considers adversaries who would try to predict or reproduce identity tokens generated by the system, assuming they have no access to unreproducible quantum randomness (in practice simulated on AWS SV1) and cannot break ToyLWE-based post-quantum signatures.
Data comprises real user interaction messages submitted via Telegram group chat messages mentioning the QSignAI bot (@mention). Message text and optional photos are ingested and stored securely in AWS-managed DynamoDB and S3 stores. Each message is associated with a unique quantum-randomness-seeded identity signature.
Two quantum circuits generate the cornerstone entropy sources:
- Circuit A: 4 qubits initialized with Hadamard gates to create superpositions, entangled with a chain of CNOTs and personalized user-seeded Ry rotations with angles derived from username characters. The circuit is executed for 100 shots on AWS Braket SV1 simulator, extracting the top measurement bitstring mapped to an integer qnum in [0,1000].
- Circuit B: 2 qubits create a canonical Bell state |Φ+⟩ via Hadamard and CNOT gates, measured over 200 shots producing the probability distribution β over basis states, used both as entropy and to map entanglement quality to visual color saturation and brightness.
The quantum outputs seed a ToyLWE-derived signature scheme using SHAKE-256, hashing the username seed, quantum number qnum, and fresh 32-byte nonce r. The resulting signatures are truncated to form a unique public key fingerprint and a 24-character base64 signature string. This is a demonstrative instantiation, with plans to replace it with production-grade CRYSTALS-Dilithium post-quantum signatures.
The system architecture layers include:
- Participation layer: Telegram messaging platform with webhook integration, parsing @mentions and handling message and photo uploads.
- Bot/API layer: Runs on AWS ECS Fargate containers, performing webhook validation, mention detection, message sanitization, and initiating asynchronous quantum pipeline tasks.
- Quantum layer: Submits circuits to AWS Braket SV1 state-vector simulator, extracts measurement results for signature generation.
- Data layer: Stores messages, badges, quantum results, and photos in DynamoDB and S3 with encryption and audit trails.
- Presentation layer: Next.js app served through CloudFront CDN, continuously polling every 5 seconds to update the public photo wall with message cards including quantum-generated badges.
Training is not applicable as this is a system integration and deployment demonstration rather than a machine learning model. However, quantum circuits were executed with fixed shot counts (100 and 200) and user-derived Ry rotation angles.
Evaluation is primarily qualitative, demonstrating latency compatibility with live event interaction (quantum tasks run asynchronously in 4-16 s, user messages and cards appear instantaneously, badges update at next poll ≤ 5 s). The system includes graceful degradation methods if quantum tasks timeout after 30 seconds, switching to cryptographic fallback without breaking UX continuity.
No formal statistical testing or adversarial evaluations are presented; measurable latency benchmarks, NIST randomness assessments, user studies on quantum literacy impact, and formal security analyses of ToyLWE signatures versus standard PQC are explicitly deferred to future work.
Code and system architecture are openly released on GitHub. Data is non-public due to privacy considerations. The quantum module currently targets AWS Braket SV1 classical simulator but is designed to be QPU-ready by swapping device ARNs.
Technical innovations
- Integration of real-time quantum-randomness generation from cloud-based quantum circuits into a high-throughput AI bot-enabled social messaging platform with acceptable latency.
- A novel two-step quantum pipeline using a 4-qubit RNG circuit combined with a 2-qubit Bell state measurement to seed identity signatures with both quantum randomness and entanglement-derived entropy.
- Encoding quantum measurement outcomes as visual identity badges via perceptually spaced HSL color mapping, making abstract quantum phenomena legible to non-specialist audiences in live events.
- A graceful degradation mechanism that seamlessly falls back from quantum randomness to local cryptographically seeded randomness on quantum task failure or timeout, maintaining continuous UX.
Datasets
- Telegram chat messages and photos — size not explicitly reported — from live event participants via QSignAI deployment (data non-public due to privacy).
- Quantum measurement results from AWS Braket SV1 classical simulator (100 shots for Circuit A, 200 shots for Circuit B) per participant.
Baselines vs proposed
- Quantum-randomness-seeded tokens vs standard PRNG-based tokens: measurable NIST SP 800-90B cryptographic quality comparisons deferred to future work.
- AWS Braket SV1 simulator latency: 2–8 seconds per circuit; asynchronous integration avoids user-visible blocking; no comparative QPU latency benchmarks performed yet.
- Fallback local cryptographically seeded signatures latency: less than quantum task timeout (30 s), ensuring UX continuity; no explicit latency values given.
Limitations
- Quantum randomness is currently simulated via AWS Braket SV1 state-vector simulator, which ultimately uses pseudo-random processes at the hardware level — true quantum randomness requires physical QPU integration.
- No quantitative evaluation of entropy quality or cryptographic strength versus classical PRNG or post-quantum signature standards is provided; NIST randomness and security benchmarks are deferred.
- No controlled user studies measuring the impact of quantum-derived visual badges on quantum literacy or engagement were conducted; these are future work.
- No adversarial evaluation or formal threat modeling against realistic attackers beyond basic assumptions.
- The ToyLWE signature instantiation is pedagogical and not production-grade; replacement with CRYSTALS-Dilithium is planned but not demonstrated.
- Dataset size and diversity details for user interaction data are not disclosed, limiting reproducibility and generalizability assessments.
Open questions / follow-ons
- How does quantum-randomness-seeded identity token quality and unpredictability compare to classical PRNG tokens across standard cryptographic randomness benchmarks (e.g. NIST SP 800-90B)?
- What are the performance and latency tradeoffs when running the quantum circuits on physical QPUs (IonQ, Rigetti, IBM) compared to simulators?
- Can exposure to quantum-derived visual badges measurably improve quantum literacy or engagement in controlled user studies?
- How do ToyLWE-based signatures seeded by quantum randomness compare in security and efficiency to standardized post-quantum signature schemes like CRYSTALS-Dilithium in production?
Why it matters for bot defense
For bot-defense and CAPTCHA practitioners, QSignAI introduces a new paradigm where identity tokens can be seeded with cryptographically irreproducible quantum randomness, moving beyond deterministic pseudo-random generation common in existing systems. This reduces risks of token replay or prediction by adversaries, enhancing identity robustness. The integration with a widely adopted messaging platform and a conversational AI interface demonstrates feasible real-time use with manageable latency and graceful degradation fallback mechanisms to preserve availability.
While still experimental and deployed on quantum simulators, the architecture is QPU-ready, pointing to future opportunities for embedding provably quantum-secure randomness in bot-defense contexts. Additionally, visual badges encoding quantum entanglement states as color cues provide a novel avenue for raising user awareness or tying identities to quantum provenance transparently, potentially informing next-generation human verifications or CAPTCHA designs leveraging quantum principles. However, practitioners should note the lack of current adversarial or distribution shift testing and the reliance on simulated rather than physical quantum randomness in this initial work.
Cite
@article{arxiv2605_27729,
title={ QSignAI: Quantum-Randomness-Seeded Identity Signatures at the Intersection of AI for Science and Science for AI },
author={ Dongping Liu and Aoyu Zhang and Luyao Zhang },
journal={arXiv preprint arXiv:2605.27729},
year={ 2026 },
url={https://arxiv.org/abs/2605.27729}
}