Towards the Detection of Diffusion Model Deepfakes