Security experts are warning that a security flaw has been found in a popular and powerful data encryption algorithm, dubbed SHA-1, by a team of scientists from Shandong University in China. The three scientists are circulating a paper within the cryptographic research community that describes successful tests of a technique that could greatly reduce the speed with which SHA-1 could be compromised.

Although the cracking technique could not be carried out practically, it does compromise the integrity of the algorithm and could lead to more advanced attacks that would render SHA-1 useless, affecting many Internet security products that use it to generate digital signatures, according to Bruce Schneier, founder and CTO of Counterpane Internet Security.

SHA-1 is a popular encryption algorithm that was developed by the U.S. National Security Agency (NSA) in 1995 after a weakness was discovered in a predecessor algorithm, called the Secure Hash Algorithm, or "SHA." The algorithm is among those most commonly used to generate "hashes," or unique strings of values that are used to encrypt and decrypt digital signatures, Schneier said.

SHA-1 is used to create signatures by most of the popular security protocols on the Internet, including SSL and PGP (Pretty Good Privacy), he said.

A research team of three scientists: Xiaoyun Wang, Yiqun Lisa Yin, and Hongbo Yu, is circulating a paper called Collision Search Attacks on SHA-1 that describes methods for creating so-called "collisions" with the SHA-1 algorithm 2,000 times more quickly than had been possible before.

"It's phenomenal research," Schneier said. "There's a lot of really impressive math."

A "collision" is an occurrence in which two messages have an identical hash value. It opens the door to forging valid signatures generated using SHA-1. Cryptographers rely on "non repudiation" in algorithms, the concept that two identical hash signatures cannot be created by different signers, said Michael Szydlo, a senior research scientist at RSA Security's RSA Labs.

The results of the paper mark a significant improvement over previous methods of cracking SHA-1 but still require a massive number of attempts to work -- a number expressed by 1 with thirty zeros after it, he said.

The IDG News Service is a Network World affiliate.