A new attack against RSA encryption has been demonstrated that uses a different approach than the factoring methods long assumed to be RSA’s primary vulnerability. The technique, which enables signature forgery, was successfully applied to deprecated 1024-bit RSA keys using an academic computing cluster over several months—substantially less effort than factoring would require.
The attack implements a variant of the number field sieve algorithm through an oracle-based approach that gathers information from multiple queried inputs. Against 1024-bit keys, the method requires approximately 2^65 operations and roughly 1,380 core-years of computation, compared to the estimated 2^80 operations and 500,000 to 1 million core-years needed for traditional factoring. For 2048-bit and 4096-bit keys, the security levels drop to 2^90 and 2^119 respectively, all below the minimum 128-bit threshold required by security agencies including the NSA and NIST.
The practical threat remains limited to specific implementations. The attack only affects blind-signature or textbook RSA, not the padded variants like PKCS or PSS that protect the majority of RSA deployments in use today. However, some real-world systems continue to use unpadded RSA, most notably Privacy Pass, a privacy authentication protocol deployed by Apple and Cloudflare. An attack on those systems would require requesting approximately 2^43 tokens, a volume comparable to daily network traffic at major providers.
Cryptography experts have characterized the discovery as conceptually significant despite its limited immediate impact. Researchers note that the hand-coded implementation did not utilize AI or GPU acceleration, suggesting security levels could deteriorate further with optimization. The finding is expected to accelerate efforts toward transitioning to quantum-resistant cryptographic systems, which have been pursued as quantum computing capabilities continue advancing.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI