Part V - Information Warfare and the Future of Truth — Synthetic Consensus and Agentic AI

In Parts I through IV, we traced the arc of surveillance from Richelieu’s axiom to the bureaucratic violence of zombie data. We met the Aurora family, forced to the pavement at gunpoint by an ALPR error. We met Robert Williams, arrested in his driveway by a facial recognition algorithm. We met Rafaela Aldaco, denied housing by a zombie record from a dismissed charge. We documented the 19.5 percent drop in Wikipedia traffic that proves surveillance silences the innocent. Throughout, we assumed a flawed but functional information environment—one where the truth exists, even if the system distorts it. In this final installment, we confront the possibility that the truth itself is being manufactured.

5.1 The Rise of Cyber Troops

While citizens are chilled into silence, state actors are ramping up the volume of manufactured noise. The Oxford Internet Institute has documented the global rise of “cyber troops”—government or political party teams tasked with manipulating public opinion through social media. By 2019, such operations were active in seventy countries.⁶⁴ These troops use “computational propaganda” to flood the zone with noise, drowning out organic dissent and making it impossible for ordinary citizens to distinguish authentic public opinion from state-manufactured consensus.

5.2 Agentic AI Swarms: The Next Threat

The evolution beyond cyber troops is the Agentic AI Swarm. Unlike traditional botnets, which are often clumsy and easily detected, agentic AI swarms utilize Large Language Models (LLMs) to create autonomous agents that maintain persistent identities, memory, and distinct personalities.⁶⁶

These swarms can coordinate to manufacture “synthetic consensus”—the illusion that a specific viewpoint is widely held. A malicious swarm can infiltrate an online community, adapt its tone to local norms, and gradually shift the “Overton window” of acceptable discourse without any human oversight.⁶⁶ This capability threatens to break what scholars call the “epistemic substrate” of democracy. If eighty percent of the voices in a public debate are AI agents feigning human outrage, actual humans will conform to the synthetic majority—believing it to be the real one.

5.3 LLM Grooming vs. Data Voids: How Disinformation Enters AI

A contentious scholarly debate has emerged regarding how disinformation enters the AI models that increasingly mediate public knowledge.

One camp argues for “LLM grooming”: the theory that adversaries—Russian operatives among them—are deliberately flooding the web with disinformation to poison the training data of future AI models, ensuring that chatbots regurgitate propaganda as though it were fact. This is also termed “data poisoning.”⁶⁸

A second camp offers a more structural explanation: “data voids.” When chatbots cite disinformation, it is often because high-quality information on a topic simply does not exist in sufficient volume, and the void is filled by the only available sources—which happen to be propaganda.⁷⁰ If a conspiracy theory is not debunked by mainstream sources (creating a void), the LLM retrieves the conspiracy content because it is the only relevant text available.

Both explanations may be simultaneously true—and both point to the same conclusion: the information ecosystem on which AI depends is vulnerable to corruption, whether by design or by neglect.

5.4 Legislative and Technological Defenses

Combating the Digital Panopticon requires action on multiple fronts simultaneously. The architecture of surveillance was built over decades; dismantling it will require sustained legislative, technological, and cultural effort.

Legislative Reforms

The Fourth Amendment Is Not For Sale Act would prohibit law enforcement from purchasing data from brokers that they would otherwise need a warrant to obtain, closing the “checkbook surveillance” loophole that has allowed agencies to bypass judicial oversight entirely.⁷²

The American Privacy Rights Act (APRA), a proposed comprehensive federal privacy law, would establish data minimization standards and create a registry of data brokers—bringing transparency to an industry that has operated largely in the shadows.⁷⁴

At the state level, proposals for “Statutes of Erasure” would mandate the deletion of ALPR data after defined periods—Maine, for example, requires deletion after 21 days—and compel the expungement of zombie records from private databases.⁷⁶

Technological Defenses

On the technology side, systems like “Vanish” propose mechanisms by which data automatically becomes unreadable after a set period, achieved by destroying the encryption keys distributed across a network.⁷⁷ The growing adoption of disappearing-message applications like Signal represents a grassroots defensive measure against the “infinite memory” of the cloud.⁷⁹

Neither legislation nor technology alone will suffice. What is needed is a cultural shift: a recognition that the “data retention by default” model is not merely inefficient or risky, but fundamentally incompatible with the kind of society most people say they want to live in.

5.5 Conclusion: Reclaiming the Right to Be Forgotten

Over the course of this series, we have given you the evidence. The trajectory of the twenty-first century has been the construction of a perfect surveillance machine, fulfilling Richelieu’s dream of total visibility. We have given the state not six lines, but six billion. The consequences are a chilling of dissent, a rise in wrongful algorithmic accusations, the quiet violence of administrative databases, and the manufactured corruption of reality by synthetic agents.

The remedy requires a fundamental rejection of the premise that more data always means more safety. It requires laws that forbid the purchase of our private lives, technologies that ensure our digital pasts can die, and a societal recognition that “perfect memory” is incompatible with human liberty. The ability to forget—to move past mistakes, to grow, to begin again—is not a bug in the human condition. It is the feature that makes freedom possible.

Cardinal Richelieu believed that six lines could hang any man. He was limited by the technology of his century—ink, paper, informants with finite memories. We have removed every limitation he faced. The archive is infinite. The readers are algorithms. The memory is permanent. And the lines we write—our searches, our movements, our faces, our associations—are no longer six. They are six billion, generated every hour, stored forever, and increasingly interpreted by machines that cannot distinguish the innocent from the guilty because they were never designed to.

Unless we establish a “right to be forgotten” and a “right to be unobserved,” we will remain inmates in the Digital Panopticon—forever writing the lines that will one day be used to hang us.

•  •  •

Table 1: Comparative Analysis of Surveillance Mechanisms

Works Cited

8. “26 Billion Records Released in the Mother of All Breaches,” McAfee Blog, accessed January 30, 2026

64. “Troops, Trolls and Troublemakers: A Global Inventory of Organized Social Media Manipulation,” Oxford Internet Institute, accessed January 30, 2026

66. Schroeder et al., “How Malicious AI Swarms Can Threaten Democracy,” arXiv:2506.06299, accessed January 30, 2026

67. “AI swarms could fake public consensus and quietly distort democracy,” City St George’s, University of London, accessed January 30, 2026

68. “LLMs grooming or data voids?” HKS Misinformation Review, accessed January 30, 2026

70. “LLMs grooming or data voids? LLM-powered chatbot references to Kremlin disinformation,” HKS Misinformation Review, accessed January 30, 2026

72. H.R.4639, “Fourth Amendment Is Not For Sale Act,” 118th Congress, accessed January 30, 2026

74. “The American Privacy Rights Act,” Congress.gov / CRS, accessed January 30, 2026

76. “An Act To Regulate the Use of Automated License Plate Recognition Systems,” Maine Legislature, accessed January 30, 2026

77. “SafeVanish: An Improved Data Self-Destruction for Protecting Data Privacy,” IEEE Xplore, accessed January 30, 2026

79. “Ediscovery and Ephemeral Messaging Data: The Complete Guide,” DISCO, accessed January 30, 2026