One of the biggest frustrations I have with tech blogs are the fake products that are too good to be true. Feeds are constantly riddled with fake product hacks that are mocked up or modified by photoshop experts. See mini-iPhone.
To the human eye, it is very difficult to determine whether or not an image is fake. However, there are some methods for exploiting the photoshop techniques through image anaylsis, particularily the clone stamp tool. For those not familiar with the stamp tool, it basically involves copying a region of an image to different regions in order to realistically mask objects or fill empty space. See link. The efficient technique for exploiting these photoshop flaws is rather simple. An image is divided into a number of blocks that are larger than the pixel size yet smaller than the duplicated region. By first applying principal component analysis (PCA) on these image blocks, a lexigraphically sorted matrix with reduced dimensions is created: Using least squares the original image can be represented in fewer dimensions while preserving the most important contrasting components; very important for efficiency. The matrix is then sought through to find groups of image blocks with relatively the same characteristics as blocks located elsewhere in the matrix, thus finding a duplicated region.
Although this method will only detect usage of the clone tool, it is a step closer to being able to effectively detect image forgeries.

The algorithm: http://www.ists.dartmouth.edu/library/tr-2004-515.pdf