According to a report in the Oct. 2 issue of Nature, IP Australia is taking some heat for issuing a patent (AU2004309300B B2) to discredited Korean scientist Dr. Woo Suk Hwang that was based on fabricated data. In 2004 and 2005 Dr. Hwang claimed to have created a stem-cell line from a cloned human embryo. Hwang's claims, it was discovered, were a complete fraud. He lost his university job, was convicted of various crimes including embezzlement and his published papers were retracted. However, his university continued to prosecute patent applications based on his work.
Unlike scientific papers, patents are not peer-reviewed. Patent offices issue patents based on the legal criteria put for under national patent laws. They don't test the validity of the data submitted with a patent application. If they did, the patent system would probably grind to a halt. Of course, there are penalties for inventors who lie or commit misrepresentations in a patent application. But this usually applies to things like the date of conception of the invention, true inventorship, etc.