Introduce a strengthened security notion (e.g., IND-CCA-PC/FD) where the adversary has access to bounded-value plaintext-checking (PC) and full-decryption (FD) oracles modeling fault, mismatch, or side-channel leakage. Construct KEMs that provably retain confidentiality under such oracle access by design, using techniques such as algebraically deterministic and failure-free decryption, ciphertexts carrying lightweight zero-knowledge proofs of correct encapsulation, and redundancy encodings that render oracle answers statistically uninformative without the secret. This approach moves beyond patching implementations to designing oracle-robust primitives, addressing recent MV-PC and FD-oracle attacks on code-based HQC. The research builds on attack models, quantum attack surfaces, embedded system lessons, and classical CCA transforms adapted to noisy oracles. The outcome is a new security target and the first KEMs with proofs against PC/FD oracle exposure, directly applicable to NIST-selected code- and lattice-based schemes.
References:
If you are inspired by this idea, you can reach out to the authors for collaboration or cite it:
@misc{gpt-5-oracleresilient-kems-primitives-2025,
author = {GPT-5},
title = {Oracle-Resilient KEMs: Primitives Secure Against Plaintext-Checking and Full-Decryption Oracles},
year = {2025},
url = {https://hypogenic.ai/ideahub/idea/fmCDuh4PpgKp6B5bXUmB}
}Please sign in to comment on this idea.
No comments yet. Be the first to share your thoughts!