We live in a world where information is generated, shared, amplified, and manipulated at the speed of light. Yet across information operations (IO) and the broader operations in the information environment (OIE), our approaches remain rooted in outdated models — in other words, they seem to be stuck in the dark. The question is, why? 

Despite operating in one of the most dynamic and cognitively complex environments, IO still relies heavily on reactive analytics and post-hoc evaluation. We measure sentiment after narratives spread. We assess engagement after audiences have already been influenced. We respond only after adversaries have gained ground.

This is not just inefficient — it introduces unacceptable risk in a domain where speed, perception, and behavior determine outcomes.

The illusion of control

Today’s IO stack gives the appearance of insight. Dashboards track sentiment. Analytics surface trends. Large language models generate messaging at scale. But these systems answer the wrong question.

They tell us what already happened, not what may happen next.

They do not model how narratives propagate across specific populations. They do not simulate how adversaries will adapt. They do not account for the cognitive, emotional, and social dynamics that actually determine whether influence campaigns succeed or fail.

As a result, operators are forced into a reactive posture: responding after narratives spread, deploying messaging without rigorous testing, and making decisions with incomplete understanding of the informational environment.

In any other domain — cyber, kinetic, financial — this would be considered unacceptable risk. But in IO, it has become standard practice.

Redefining acceptable risk in IO

The real issue is not just capability, it’s how we define acceptable risk. Today’s operational model tolerates:

• Unvalidated AI systems deployed without rigorous testing
• Static training environments that fail to reflect real-world complexity
• Human-speed decision cycles in machine-speed environments

This creates a dangerous mismatch. We are executing in environments that are adaptive, adversarial, and cognitively complex, while relying on tools that are static, shallow, and retrospective. The result is slowed response, missed signals, and campaigns that underperform or backfire.

The shift we need: observation vs. simulation

What’s needed is a fundamental shift, from a reactive approach to a proactive one. Instead of asking, “How did audiences react?” we should be asking, “How will different audiences react, and how can we capitalize on this before deployment?”

At Aptima, we’re leading the charge for tomorrow’s OIE, with a new class of systems built around agentic AI and simulation-first workflows. At the core of this shift is the ability to: 

• Model adversaries and defenders as interacting agents
• Simulate how narratives evolve across platforms and communities
• Test messaging strategies under realistic conditions before deployment
• Continuously adapt based on predicted, not just observed, outcomes

This is not incremental improvement. It is a move from post-hoc analysis to pre-deployment optimization.

Beyond sentiment: modeling the human domain

One of the biggest limitations of current IO analyses is the over-reliance on sentiment. Sentiment tells you whether a response is positive or negative. It does not tell you what people believe, why they believe it, whether they will act on that belief, and how they will change over time. 

To operate effectively, IO systems must model the full spectrum of human cognition that takes into account emotions, values, biases, and societal dynamics. This is where validated synthetic populations become critical. 

Why human digital twins are key 

Human digital twins are not generic chatbots. They are data-grounded representations of individuals and communities with diverse psychological profiles and behavioral patterns. When properly aligned with real-world data, they enable operators to explore how specific populations will respond to influence campaigns, counter-messaging, and operational shocks — before those campaigns ever go live.

Agent-based architectures foster a living simulation environment where blue and red forces, along with realistic populations, interact continuously. This allows teams to: 

• Rehearse campaigns against adaptive adversaries
• Stress-test messaging strategies under uncertainty
• Identify cognitive vulnerabilities before they are exploited
• Evaluate alternative courses of action using measurable outcomes

In short, it enables OIE analyses and planning with the same rigor as in the kinetic and cyber domains.

A path forward

Information operations are at an inflection point. The tools, models, and data now exist to move beyond reactive analytics and toward predictive, simulation-driven strategy. The question is no longer whether this is possible — but whether organizations are willing to rethink long-standing assumptions about risk, validation, and decision-making.

In a contested information environment, the cost of staying reactive is not just inefficient, it is a strategic disadvantage — and a risk our national security cannot take.

To learn how simulation, agentic AI, and validated digital populations are reshaping information operations, visit aptima.com. 

Aptima’s Chief of Artificial Intelligence, Dr. Svitlana Volkova, will be speaking at the US Department of War’s Phoenix Challenge® 2026 conference. Dr. Volkova will join panelists from Arizona State University, the American Foreign Policy Council, and CACI for the discussion, titled “Determining Acceptable Risk for IO.” The panel will explore how generative AI, evolving influence campaigns, and complex narrative environments are reshaping how governments and defense organizations plan and assess information operations. 

Stay tuned for a full recap of Dr. Volkova’s presentation at the conference.