A recent weakness has been specifically identified in the Android SecureRandom implementation. Oracle JRE and OpenJDK have no vulnerabilities in their SecureRandom implementations, which are still unknown, as far as I can tell. It is worth noting that the implementation of SecureRandom will differ depending on where the JVM works (that is, Windows versus Unix-like)
EDIT: Excerpt from the link to the article you added, emphasizing mine.
The problem is that Androids has a built - in pseudo - random number generator, the SecureRandom Java class.