AI Propaganda Overwhelms Democratic Verification
A Vanderbilt University paper indicates the US Department of Defense accounts for 70-90% of federal AI contracts. Meanwhile, Forethought.org reported that 74.5% of fact-checking organizations operated on budgets under $500,000 in 2025, with 14.2% on less than $20,000. The Stimson Center observes that as citizens and experts struggle to distinguish AI-generated falsehoods from genuine information, cynicism, news avoidance, and social disengagement increase. A Colorado State University analysis pointed out that when experts and news agencies fall for AI hallucinations, public trust in institutions like journalism and electoral processes suffers further. The Brookings Institution, Media Engagement, the Brennan Center for Justice, and the National Endowment for Democracy have documented how the "liar's dividend" allows malicious actors to dismiss authentic evidence as AI-generated fabrications, evading accountability. Conversely, a USC Viterbi School of Engineering study found that state actors use this volume to streamline operations, deploying AI agents that autonomously coordinate, write posts, and echo content to manufacture consensus.
State Computational Advantage Entrenches Power
Media Engagement found that wealthier actors and states are more likely to implement generative AI effectively, further entrenching existing power dynamics. While AI democratizes high-fidelity content generation, a compounding computational and data gap amplifies state resource advantages and entrenches state power, Media Engagement reported. A USC Viterbi School of Engineering study stated that AI agents can autonomously coordinate, write posts, learn from teammates, and amplify content without human direction, manufacturing seemingly genuine discussions. The National Endowment for Democracy and the Middle East Policy Council detailed how, beyond sheer volume, states utilize a data advantage through mass surveillance and biometric collection to fuel "precision cognitive attacks." The National Endowment for Democracy explained that these systems target individuals based on personalized user portraits derived from scraped social media data, enabling hyper-tailored influence operations. For instance, the Middle East Policy Council reported that Israel uses advanced AI systems to collect biometric data and iris scans from over 2 million Palestinians in Gaza and the West Bank to predict movements and facilitate arrests.
Experts Fail to Detect Deepfakes
The Colorado State University analysis and the Stimson Center reported that despite these efforts, human evaluators and experts perform little better than chance at detecting deepfakes. The Colorado State University analysis, the Brookings Institution, the Stimson Center, the Middle East Policy Council, and the Carnegie Endowment for International Peace assert that structural accountability in democratic information ecosystems relies on verifiable content provenance, institutional transparency, and distributed public literacy. However, the shift to AI-generated propaganda significantly challenges content provenance and institutional transparency. The Brookings Institution, the Stimson Center, and the Carnegie Endowment for International Peace noted that technical standards and watermarking, such as those mandated by the EU AI Act or developed by tech companies, are often imperfect, easily bypassed, and quickly rendered ineffective by the rapid evolution of generative models. The Brookings Institution found that tech companies have severely limited researcher access to social media data, hindering the ability to pinpoint information operations and assess their scope. The Colorado State University analysis, the Stimson Center, and the Carnegie Endowment for International Peace contend that distributed public literacy, encompassing preventative education and "prebunking" campaigns that educate people about disinformation tactics before they encounter false information, directly counters the "fog of information" by empowering citizens to discern falsehoods. NewsGuard found that leading AI chatbots spread false information 35% of the time when prompted with controversial topics.
October 2023 Slovak Deepfake Audio Went Viral
The National Endowment for Democracy reported that a Russian-backed network produced over one million German-language posts, overwhelming civic verification efforts. The historical track record of early digital information operations reveals that civic counter-strategies struggle to scale against exponential volume and speed, and current AI shifts amplify pre-existing informational vulnerabilities while fundamentally restructuring power dynamics. A USC Viterbi School of Engineering study indicates that while early bot campaigns relied on scripted behaviors, current AI agents autonomously coordinate and generate content. The National Endowment for Democracy reported that this overwhelms civic verification, citing the Russian-backed network's production of over one million German-language posts. Morgan Wack, Carl Ehrett, Darren Linvill, and Patrick Warren, writing in PNAS Nexus, explained that AI amplifies existing issues by increasing the diversity of topics covered, citing a state-affiliated site that nearly doubled its topic entropy after adopting AI. The National Endowment for Democracy also noted that AI fundamentally restructures power by enabling "precision cognitive attacks" that target individuals based on personalized user portraits derived from scraped social media data. The Carnegie Endowment for International Peace documented a 2020 U.S. state legislator experiment where approximately 35,000 AI-generated emails were statistically indistinguishable from human-written ones in eliciting responses, demonstrating a failure to discern machine-generated communications.
Low-Cost State Provocations Force High-Cost Responses
This economic asymmetry means low-cost digital provocations from state actors force high-cost institutional responses from civic entities. The shift to AI-generated propaganda implies a continuous, data-driven feedback loop where state actors use superior resources for autonomous scale and precision targeting, creating a structural disadvantage for civic institutions, which face severe computational and budgetary disparities.
Comments ()