Classical propaganda aims to convince. It advances a message, targets an audience, and seeks belief or compliance. Much of the public debate about AI-driven information operations still assumes this model. That assumption is increasingly wrong.

At state scale, the dominant use of generative AI is not persuasion, but flooding. The objective is not to make people believe a particular claim, but to exhaust their capacity to determine what is worth believing at all.

This shift marks a qualitative change in how information environments are contested.

The core mechanism: epistemic exhaustion through volume

Generative AI makes it possible to produce and distribute enormous quantities of plausible content at negligible cost. When deployed in information operations, this capability is not used to craft a single compelling narrative, but to saturate the environment with many overlapping, inconsistent, or redundant ones.

The mechanism is simple. When volume exceeds verification capacity, audiences stop checking. They disengage, retreat to prior beliefs, or rely on identity cues rather than evidence. Confusion becomes the outcome, not persuasion.

This is not a failure of public reasoning. It is a rational response to overload.

How flooding differs from persuasion

Persuasion is directional. It depends on coherence and repetition. Flooding is ambient. It depends on noise.

In a flooded environment:

  • contradictions do not weaken impact,
  • low-quality material is sufficient,
  • and attribution is optional.

What matters is that every channel feels unreliable. When official statements, journalism, eyewitness media, and synthetic content blend together, trust collapses across the board.

Concrete state-level examples

This pattern has already been documented in multiple geopolitical contexts.

During the war in Ukraine, analysts observed waves of AI-assisted content—text, images, and videos—circulating across platforms, often rapidly produced and lightly edited. The content was not consistently persuasive or even internally coherent. Its effect was to muddy timelines, overwhelm fact-checking capacity, and create uncertainty about what could be trusted.

Similarly, investigations into Chinese and Iranian information operations in the mid-2020s noted a shift toward high-volume, low-distinctiveness messaging, often amplified through automated accounts. The goal was not to win arguments, but to degrade the informational environment in which arguments occur.

These efforts succeeded not by convincing skeptics, but by making engagement feel futile.

Why attribution collapses at scale

Flooding operations exploit a critical weakness in institutional response: attribution.

When content volume is low, institutions can ask who produced it, why, and under what authority. When volume is high and content is recombinable, attribution becomes slow, uncertain, and politically contested. By the time attribution is established, the moment has passed.

This creates a structural asymmetry. States deploying flooding tactics face little downside, while states attempting to respond must adhere to evidentiary and legal standards that slow reaction.

Why counter-messaging and literacy fail

Traditional counter-propaganda tools assume a persuadable audience and a limited set of false claims. Flooding defeats both assumptions.

Counter-messaging adds to volume. Media literacy helps individuals assess claims but does not reduce the ambient noise they must navigate. In a saturated environment, even accurate information struggles to surface.

The problem is not that people cannot tell truth from falsehood. It is that they no longer expect to be able to.

A concrete institutional marker

The European Union’s public assessments of foreign information manipulation in the mid-2020s explicitly acknowledged this shift, noting that many operations aimed less at narrative dominance than at destabilizing trust in institutions and media. The diagnosis marked an important recognition: the harm lay in environmental degradation, not message success.

That recognition, however, has not yet translated into effective containment.

Why this belongs in the dark patterns arc

Flooding is a dark pattern at the societal level because it exploits the asymmetry between cheap disruption and expensive verification. It leverages open information systems against themselves, without requiring deception in the narrow sense.

Once flooding becomes normalized, restoring trust is far harder than preventing its erosion. Information environments degrade faster than institutions can adapt.

What is missing, institutionally

From an ACP perspective, this domain lacks:

  • enforceable constraints on synthetic amplification,
  • durable provenance mechanisms that survive recombination,
  • and shared norms for throttling volume in contested environments.

Absent these, states face a choice between tolerating epistemic erosion or imposing blunt controls that undermine open discourse. Neither option is attractive, but the space for intermediate solutions narrows over time.

The danger is not that propaganda becomes more effective. It is that information itself becomes less usable as a basis for collective action.