RSA public keys contain a modulus value N that is the product of two primes p and q. The security of RSA relies on the fact that p and q are secret and factoring N is a difficult problem and not computationally feasible. With knowledge of p and q the private key can be trivially calculated.
When two RSA keys share one of their prime factors then this shared prime factor can be calculated efficiently with the greatest common divisor (GCD) algorithm. There also exists an efficient algorithm to search for common prime factors in a large set of keys.
In 2012 two research teams independently applied this method to large collections of public keys. The analysis by one team indicated that these keys were mostly generated due to early boot time entropoy problems on Linux.
In 2016 a research team published an updated analysis coming to the conclusion that this vulnerability remains widespread. Among TLS keys this vulnerability remains one of the most common flaws found in the wild.
We detect affected keys by having a collection of known GCDs. By calculating a GCD against a product of all known GCDs of a fitting size we are able to efficiently detect affected keys. However we can not be sure that we cover all vulnerable GCDs.