Can we trust what we see? Although AI has become clever, there are still small but significant mistakes to identify deepfakes 2 hours ago