Steganalytic techniques are used to detect whether an image contains a hidden message. By analyzing different image features between stego-images (the image within which information is hidden) and cover-images (the image in which information is to be hidden), a steganalytic system is able to detect stego-images. In this paper, we present a new method in LSB embedding by avoiding the change of statistic features. After embedding data in the first LSB bits, we apply genetic algorithm by adjusting the second LSB bits of a stego-image while creating the desired statistic features to generate the modified stego-images that can break the inspection of steganalytic systems. Experimental results show that our algorithm can not only pass the detection of chi-square test, but can also leave our hidden message unchanged on the first LSB bit, and enhances the peak signal-to-noise ratio of stego-images.