The Arkansas Center for Data Sciences (ACDS) recently published a guest column by COSMOS director Dr. Nitin Agarwal. ACDS is a nonprofit that seeks to support IT professionals and further industry knowledge. For example, ACDS organizes data science events, employs apprenticeships and internships, and helps IT employers and employees find each other.  In his guest column, Dr. Agarwal addresses the danger of deepfakes and AI-amplified propaganda, disinformation, and misinformation. As he describes in one excerpt, “Let’s say someone has an image of your face plus samples of your voice from your public appearances and speeches. Those speeches can be used to train the model to speak a new script, in your voice and with your inflections, and the video can be displayed on top of that newly created voice. It appears now like you have said those very words, which you have not.”

Dr. Agarwal explains that one solution to combat deepfakes and general AI-amplified propaganda is to “develop more media literacy and more critical thinking skills.” However, Dr. Agarwal also highlights that the same AI technology that creates deepfakes can be further repurposed to identify and combat them, i.e., fight fire with fire. “We must simultaneously use AI capabilities to detect how they’ve been created and whether a particular image is a true image versus generated, a fake, or a deepfake,” Dr. Agarwal says. “We can measure how much the deepfake has been circulating, identify who has shared it, who has engaged with it, and where focused intervention efforts (e.g., inoculation, education) are needed.” 

But to create practical methods to combat propaganda requires research, especially when it comes to addressing and creating new AI and technology. Dr. Agarwal illustrates, “Theoretical grounding plus computational methodology provides a robust solution to this problem.” He finishes by emphasizing that COSMOS is where “we are developing research-driven solutions validated by real-world data.” In other words, COSMOS helps connect and bridge the gap between theory and practical methodology, creating capabilities for policymakers, warfighters, and the general public to identify and combat propaganda like deepfakes.  

The full guest column written by Dr. Agarwal can be found at