Intrusion detection systems and file scanning security software are widespread tools to secure systems; however, these tools require high levels of automation, monitoring, and maintenance. Because of this, using an artificial intelligence to determine if a file is malicious is tempting due to artificial intelligence’s ability to process large amounts of data and identify hidden patterns. The goal of this research project is to determine the efficacy of using detection artificial intelligence when determining if a file is malicious or benign. This is done by training models using datasets of malicious files and benign files converted to 128 \(\,\times \,\) 128 PNG files. Initially, the models are tested for accuracy and consistency; however, the models are then attacked using well known evasion techniques. Later, the models are retrained in three different ways using the attack dataset and tested again to determine the effectiveness of the countermeasures. Previous research on the effects of attacks against artificial intelligence for other purposes has shown high impact on accuracy, and countermeasures against these attacks have also shown a pattern of limited resistance against the attacks. Due to the results of this previous research, the accuracy scores of these models are expected to be decreased when confronted with attacks, and accuracy is expected to decrease less with countermeasures.