Based on expert discussions and intelligence insights
By Benoît Grenier
Strategic Advisor — Intelligence, Risk Management & Counter-Intelligence
Ten essential questions every CEO must ask now
PART I
Leading in a World Where Truth No Longer Exists by Default
Modern executives are confronted with a transformation unlike anything seen in previous economic cycles: a decisive shift from an environment where information could be collected, verified, and trusted, to an ecosystem where data itself is increasingly unstable, contaminated, and adversarial. AI, originally viewed as a strategic accelerant for decision-making, has become both an indispensable partner and a systemic risk. The promise of clarity has collided with a reality of volatility.
The situation is simple and brutal: executives are asked to make decisions that shape capital allocation, operational stability, and global strategy in a world where informational certainty has collapsed. The discussions I had recently with friends and experts reveal the depth of this crisis. Experts in AI, intelligence, and risk management describe a digital environment overwhelmed by synthetic content, polluted information flows, degraded data sources, and proliferating artificial identities.
The core message emerging from that exchange is alarming but accurate: the very foundation of corporate decision-making—trusted information—is eroding at an accelerating pace. When leaders cannot trust the data feeding their dashboards, their forecasts, or their AI systems, the entire enterprise begins to drift into structural vulnerability.
This article examines that context and outlines how companies can restore strategic clarity. The goal is not to create fear; it is to present the factual landscape of risk, intelligence, and vulnerability in 2025 and beyond. Organizations that want to survive and outperform must recognize that they are already operating in an environment characterized by economic information warfare, geopolitical fragmentation, synthetic data contamination, and AI-driven ambiguity. Only those who integrate intelligence, validation, and deliberate skepticism into their decision-making frameworks will maintain their advantage. In other words, a real counter-intelligence corporate program!
The Collapse of Ground Truth
For more than a decade, executives operated under a simple assumption: the more data an organization collected, the better its insights would be. Big data symbolized power. Machine learning promised predictive accuracy. AI tools offered speed, automation, and optimization.
Yet the transcript reveals a disturbing inversion of that assumption. We are now living in an era where more data no longer equates to greater accuracy. Instead, data accumulation often generates more noise, more false signals, and more opportunities for adversarial manipulation.
Experts in my recent discussion describe a phenomenon known as autonomic data poisoning. It occurs when generative AI systems consume online content, transform it, reproduce it, and inject synthetic information back into the general ecosystem. Over time, this feedback loop dilutes authentic human content and saturates the information sphere with artificial patterns that models mistake for truth. One of my friends summarized it bluntly: human-generated content is being digested, diluted, and overwhelmed by synthetic information to the point where “the probability distribution of the original content disappears.”
This collapse of informational purity is not hypothetical. Oxford University demonstrated experimentally that when AI models feed on the output of other models, the result is a rapid drift into incomprehensible, fabricated patterns—a complete degradation of semantics. That outcome is not merely an academic curiosity; it is a warning for every organization relying on AI-powered analytics, monitoring tools, or forecasting systems. The models providing strategic insight may be learning from polluted, artificial sources rather than real-world human activity.
For companies, the implications are profound. Market data can be distorted by algorithmic amplification rather than genuine sentiment. Threat intelligence feeds may include fabricated accounts or artificially generated narratives. Predictive models used in supply chain planning or financial forecasting may be drawing on a corrupted dataset. In such conditions, decision-making becomes a dangerous exercise in navigating illusions that feel precise but are fundamentally unstable.
Hallucination as a Strategic Risk
LLMs present another silent danger: they hallucinate. They do not simply make mistakes; they generate seemingly authoritative statements that are entirely false. In financial operations, regulatory compliance, legal interpretation, security triage, and risk assessment, such hallucinations are unacceptable. A human analyst who guesses is negligent. An algorithm that confidently invents facts is a systemic hazard.
My friend “expert” highlighted how a hallucination rate that might seem small in a consumer setting becomes catastrophic in environments where precision is mandatory. A 99.5% “accuracy rate” in automated legal screening or financial reporting is not a success; it is a disaster. A single hallucinated interpretation could misrepresent a regulatory requirement, distort a risk exposure, or trigger the wrong operational response. And because AI expresses itself with confidence and fluency, executives may trust these invented outputs more readily than ambiguous human reports.
This is one of the most dangerous cognitive shifts now occurring in organizations: the erosion of healthy skepticism. In my recent discussion, someone noted a generational trend where younger managers accept the first answer produced by an AI tool as the correct answer, simply because it appears salient and well-phrased.
When speed and convenience meet linguistic authority, critical thinking erodes. Decisions gain velocity but lose depth, verification, and resilience.
This “first-answer bias” represents a structural threat to enterprise governance. Organizations that do not build countermeasures against AI hallucination—and against human overconfidence in AI-generated content—are exposing themselves to cascading strategic errors.
But it would be naïve to pretend that this vulnerability belongs only to machines. Humans fall into the same traps with alarming frequency. In high-pressure environments, even seasoned professionals take shortcuts: they rely on intuition instead of verification, repeat assumptions they have not tested, or project confidence they do not actually possess.
Cognitive fatigue, group dynamics, and the desire to appear decisive often lead people to “hallucinate” in their own way—by filling gaps with speculation, misremembered facts, or convenient narratives. These human distortions are quieter than algorithmic errors but no less dangerous. When organizations mistake consensus for accuracy or allow authority to substitute for evidence, the human mind becomes as much a source of unreliability as the AI systems it supervises.
The Disintegration of Attribution
Perhaps the most troubling transformation described in my recent exchange concerns attribution: the ability to determine whether a person, account, message, or data point comes from a real, identifiable human being. That foundation is dissolving rapidly.
Synthetic identities are proliferating across every digital platform. AI-generated faces, voices, résumés, comments, reviews, and social profiles are becoming indistinguishable from authentic human activity. My discussion warns explicitly that we are only a few years away from a digital world in which no executive can tell with certainty who is real and who is not.
This is not a future scenario; it is unfolding now. A corporation evaluating customer sentiment may be analyzing the output of thousands of artificial accounts. A hiring process may be reviewing CVs or video interviews produced entirely by synthetic agents. A supply chain audit may rely on vendor identities that do not actually exist. A geopolitical assessment may be based on artificially amplified narratives originating from adversarial information clusters. A security incident attributed to “a disgruntled employee” may, in reality, originate from a model-driven impersonation network.
In such an environment, the traditional OSINT model—collect, observe, interpret—is no longer viable. The open web has become an adversarial environment where genuine and synthetic signals coexist, merge, and reinforce one another. Executives relying on public signals without verification mechanisms risk grounding major decisions on fabrications.
To re-establish truth, advanced intelligence teams now rely increasingly on offline, physical-world, sensor-validated, or satellite-derived datasets. These data sources create an anchor of reality: a way to validate human presence, physical events, supply flows, or behaviour patterns that cannot be easily spoofed by AI.
Enterprises that fail to adopt similar validation strategies will progressively lose their ability to distinguish fact from fiction—a fatal condition in risk management.
Geopolitics and Economic Intelligence: The New Determinants of Corporate Vulnerability
The degradation of information integrity is occurring simultaneously with a geopolitical realignment that is transforming global business. Supply chains no longer follow economic logic but geopolitical necessity. Energy flows are being reorganized by conflict, sanctions, and alliances. The world’s major powers increasingly weaponize information, technology, and market access.
In this landscape, corporate strategy cannot ignore geopolitics. Companies without geopolitical intelligence capabilities behave like ships navigating a storm without radar. They may be highly efficient, but they are blind.
My conversation with my friends illustrated this shift vividly. One explains how some companies are pursuing an AI-enabled, hyper-localized expansion strategy that could absorb or eliminate entire types of competitors in the next five years.
That is not simply a business strategy; it is a form of economic warfare conducted through scale, data dominance, and technological superiority.
Enterprises across sectors—from retail to energy, transport, manufacturing, and finance—are exposed to similar disruptions. Conflicts in Eastern Europe, the Middle East, and the South China Sea demonstrate that geopolitical events now impact corporate operations with unprecedented speed. Maritime rerouting costs billions. Semiconductor export restrictions reshape entire industries. Sanctions destroy long-established supply chains overnight.
The result is clear: Companies are no longer operating in markets. They are operating on a geopolitical chessboard.
Modern economic intelligence programs in Universities, were created specifically to help leaders understand and respond to this reality. They train executives to anticipate market destabilization, detect adversarial influence, map competitive strategy, and integrate intelligence into the enterprise’s governance structure.
Such expertise is no longer a differentiator; it is a survival requirement.
