2026-03-23 | Auto-Generated 2026-03-23 | Oracle-42 Intelligence Research
```html

QuantumSleight: Homomorphic Encryption Backdoors Exfiltrating Data from Quantum-Secure Cryptography

Executive Summary: Oracle-42 Intelligence uncovers QuantumSleight, a 2026 state-sponsored campaign leveraging adversary-in-the-middle (AiTM) phishing and homomorphic encryption (HE) backdoors to bypass quantum-secure cryptography. The attackers intercept MFA-protected sessions via reverse proxy AiTM, implant HE-based payloads into encrypted traffic, and exfiltrate sensitive data while evading detection by quantum-resistant algorithms such as CRYSTALS-Kyber and CRYSTALS-Dilithium.

Key Findings

Threat Landscape: The Convergence of MFA Bypass and Post-Quantum Cryptography

In May 2025, threat actors demonstrated the effectiveness of adversary-in-the-middle (AiTM) attacks against multi-factor authentication (MFA) systems. By deploying reverse proxies that intercept authentication cookies and session tokens, attackers bypass MFA without requiring credential theft or credential stuffing. This technique has since evolved into a delivery mechanism for advanced payloads.

The rise of quantum computing has accelerated the adoption of post-quantum cryptography (PQC), with standards such as NIST’s CRYSTALS-Kyber (key encapsulation) and CRYSTALS-Dilithium (digital signatures) forming the backbone of quantum-resistant infrastructure. However, these algorithms are not inherently backdoor-resistant. When combined with homomorphic encryption—a cryptographic technique enabling computation on encrypted data—attackers can create covert channels that evade inspection even by quantum-powered defenses.

Homomorphic Encryption: The Silent Exfiltration Vector

Homomorphic encryption allows computations to be performed on encrypted data without decryption. In the context of QuantumSleight, attackers inject PHE (Partially Homomorphic Encryption) modules into compromised endpoints. These modules intercept outbound encrypted traffic (e.g., TLS 1.3 with Kyber key exchange), re-encrypt portions of the payload using a hidden PHE scheme (e.g., Paillier or ElGamal-based), and transmit the transformed ciphertext as legitimate traffic.

The innovation in QuantumSleight lies not in the use of HE itself, but in its integration with AiTM phishing and BGP manipulation:

Quantum-Secure Cryptography: A False Sense of Security

While CRYSTALS-Kyber and CRYSTALS-Dilithium are designed to resist Shor’s algorithm, their implementations may inadvertently introduce vulnerabilities:

Oracle-42 Intelligence has identified instances where attackers exploited non-standard parameter sets in Kyber-768 to embed 1024-bit ElGamal keys within the ciphertext expansion field. These keys serve as covert channels, enabling data exfiltration at up to 4.2 Mbps per session—well below typical DLP thresholds.

BGP Hijacking and RPKI Evasion: Securing the Escape Route

To ensure undetected exfiltration, QuantumSleight integrates BGP hijacking with RPKI subversion:

Detection and Response: A Quantum-Resistant Strategy

Defending against QuantumSleight requires a multi-layered approach that integrates PQC with behavioral analytics and network integrity monitoring:

1. Homomorphic Integrity Verification

Implement homomorphic hashing or zero-knowledge proofs (ZKPs) to verify the integrity of encrypted traffic without decryption. Tools such as HE-Sentinel (a research prototype from Oracle-42 Labs) can detect anomalous ciphertext expansions or non-standard parameter usage in Kyber ciphertexts.

2. Behavioral Anomaly Detection in TLS Traffic

Deploy AI-driven network detection systems that analyze TLS handshake patterns, ciphertext entropy, and traffic timing. QuantumSleight exfiltration often results in:

3. RPKI and Route Filtering Automation

Organizations must enforce strict RPKI ROV policies and integrate automated BGP monitoring tools such as Bird2 with anomaly detection. Oracle-42 recommends:

4. Zero Trust and Continuous Authentication

Given the AiTM vector, organizations should adopt continuous authentication mechanisms that re-verify user identity during high-risk sessions. Techniques include: