- Published on
RSA-896
- Authors
- Name
- DeepHorus
A researcher has factored RSA-896, an 896-bit RSA challenge number, on September 19, 2026, with the assistance of Claude, an AI system. The factorization breaks a record that had stood since 2009, when RSA-768 was cracked. The announcement includes the full 270-digit composite number and its two 135-digit prime factors.
RSA-896 belongs to the RSA Factoring Challenge, a set of numbers issued by RSA Laboratories in 1991 to gauge the difficulty of breaking RSA encryption. The challenge was discontinued in 2007, but the unsolved numbers have remained benchmarks for the factoring community. RSA-896 was widely considered the next major milestone after RSA-768 fell to a team led by Thorsten Kleinjung in 2009.
The composite number spans 270 decimal digits, with prime factors p and q each 135 digits long. The factors multiply together to produce the original RSA-896 modulus, confirming the result. The use of Claude suggests that AI-assisted methods played a central role in the computation.
This achievement demonstrates that 896-bit RSA keys can no longer be considered secure. While modern systems typically use 2048-bit or larger keys, the result narrows the gap between what is feasible and what is commonly deployed. It also provides a concrete data point for estimating the resources needed to attack larger key sizes.
Cryptographers are likely to view the result as significant but expected, given steady growth in computational power and algorithmic improvements. The involvement of an AI assistant marks a notable shift in how such computations may be conducted in the future. Some experts may caution that the method has not been fully disclosed, making it difficult to assess the true cost of the computation.
The factoring of RSA-896 will likely intensify calls to migrate from RSA to elliptic curve or post-quantum cryptography. It also raises the question of when RSA-1024, the next unsolved challenge number, might fall. For now, the achievement is a reminder that the security of RSA rests on the continued difficulty of factoring β a difficulty that is steadily eroding.
