With 18 of 24 AI risks crossing catastrophic probability thresholds, the ability to navigate organisational disruption is becoming a defining executive capability.
In June 2026, the MIT AI Risk Initiative published the most rigorous expert assessment of AI risk yet conducted. A three-round Delphi study of 272 international experts, drawn from academia, industry, government, and civil society across thirty-seven countries, was asked to rate twenty-four distinct AI risk domains on severity, likelihood, vulnerability, and responsibility.
The headline finding is precise and sobering. Under business as usual, experts assigned greater than ten percent probability of catastrophic outcomes to eighteen of the twenty-four risk domains over the next five years. Catastrophic was defined with specificity: more than one million deaths, more than one hundred billion dollars in financial loss, or equivalent civilisational-scale intangible harm.
The five highest severity risks were dangerous AI capabilities, competitive dynamics, weapons and cyberattacks, power centralisation, and false information. Even under a scenario where pragmatic mitigations are implemented, five risks retained catastrophic probability above ten percent. All twenty-four remained above five percent.
These are not theoretical projections. They are the aggregated probability assessments of two hundred and seventy-two domain experts, refined across three iterative rounds designed to surface genuine consensus rather than artefacts of group pressure.
For senior professionals, the implications are direct.
The environment in which they operate is highly likely to become significantly more challenging within a defined and measurable timeframe.
The disruptions heading toward their organisations are not speculative. They carry quantified probability assessments from some of the most rigorous expert consensus currently available. The consequences of those disruptions arriving without adequate preparation are not inconvenient. Several carry existential-threshold potential.
The risks are real. They are quantified. And they are capable of generating triggers that will reach every organisation simultaneously.
See: Prioritization of Risks From Artificial Intelligence, MIT, June 2026
What happens when these triggers hit systems
Quantifying the threat is not the same as understanding what it will do to the organisations it reaches.
Every organisation is a system. It has an identity, a purpose, and a set of values it exists to protect. When a trigger arrives that is radical enough to threaten any of those things, the system responds. Not because the people within it are obstructive by nature. Because coherence is what every viable system must maintain to survive. The response is structural, not personal.
This is the condition the Galileo Dilemma names with precision. When a trigger crosses the threshold at which it threatens what a system understands itself to be, the system does not evaluate it on its merits. It responds to the threat. Evidence, logic, and compelling argument address the cognitive layer. They cannot reach the level at which the system decides what it will protect.
The MIT data describes eighteen catastrophic-threshold risks capable of creating the kind of triggers that will reach organisations simultaneously. Each one is capable of crossing that threshold. Several are likely to do so together.
When that happens, the Galileo Dilemma does not arrive as a single identifiable event. It arrives as an operating condition. The executive who once encountered a genuine system-level resistance response once or twice in a career is now facing the realistic prospect of navigating several concurrently, with more inbound.
That is not a leadership challenge of the kind organisations have prepared for. It is a structural shift in what senior professional life requires.
See: The Galileo Dilemma and the Viable System Model
What poor navigation produces
The MIT study is new. The pattern it is about to trigger is not.
Thirty years of change research has documented with uncomfortable consistency what happens when organisations encounter triggers they cannot navigate. McKinsey’s analysis of change programme failure attributes the primary cause to employee resistance and lack of management support. Prosci’s research, drawn from hundreds of change practitioners across multiple studies, identifies resistance as the central obstacle in the majority of failed initiatives.
The precise figure is debated but the pattern is not. Across decades, sectors, and geographies, the research consistently points to the same conclusion. Most significant change initiatives fail to achieve their intended goals.
That figure is not a commentary on strategy or resourcing. It is a finding about capability. Specifically, the absence of the capability required to move a complex human system through resistance it generates automatically when its identity is threatened.
The cost of that absence compounds in ways most organisations do not fully account for. A failed transformation consumes significant resource. It erodes workforce confidence in leadership. It narrows the window in which necessary change can be successfully navigated. Systems learn. A failed transformation does not return the organisation to where it started. It trains the system how to resist the next attempt.
The Centre for Creative Leadership’s research on executive derailment points to the same gap from a different angle. Technical competence rarely accounts for senior leader failure at the highest levels. The inability to navigate complex human systems does so with striking regularity. The executives who derail are not typically those who lack knowledge or strategic clarity. They are those who could not move the system when the system needed to move.
Now consider what the MIT data introduces into that picture. Not one trigger. Not a single transformation initiative to be navigated carefully over a planned timeline. Eighteen catastrophic-threshold risk domains, each capable of generating triggers that produce exactly the system-level coherence response that thirty years of research says most organisations handle badly.
The organisations that have already cycled through failed change efforts are the least prepared for what is coming. They have developed sophisticated resistance while their navigation capability has stagnated. The MIT triggers will not arrive into neutral territory. They will arrive into systems already primed to resist.
Poor navigation will not merely slow the response. In organisations facing existential-threshold triggers, it will determine whether the system survives the encounter at all.
See: Why Change Initiatives Fail: The Political Dexterity Gap
What successful navigation requires
The research has circled this territory for decades without naming what it actually points toward. McKinsey speaks of leadership commitment. Prosci speaks of sponsorship and resistance management. Kotter speaks of building coalitions and creating urgency. Each captures something real. None names the underlying capability with the precision that makes deliberate development possible.
That capability is Political Dexterity.
Political Dexterity is the cognitive and implementation capability required to move a complex human system toward an outcome it would not reach on its own. It operates on two levels simultaneously. The first is cognitive. Reading the system accurately and continuously. Understanding where power actually flows, which rarely matches the organisation chart. Seeing where resistance will emerge before it arrives. Tracking how the system responds and adapts as conditions shift.
The second is implementation. Acting on what that reading reveals. Choosing the right moment. Framing necessary change in terms the system can receive without triggering its coherence mechanism. Building conditions in which resistance dissolves rather than hardens. Knowing when to push and when to allow the system’s own momentum to carry the direction forward.
Neither layer works without the other. The senior professional who reads the system accurately but cannot act on what they see achieves clarity without impact. The one who acts without reading generates effort without traction. Thirty years of change failure is largely a history of capable people doing one without the other.
But Political Dexterity alone is not sufficient for what the MIT data describes. Eighteen catastrophic-threshold risks capable of generating simultaneous triggers do not present as clean, sequenced problems. They present as conditions of genuine uncertainty, where the problem shifts as you engage with it, where every option carries credible opposition, and where the decision made today changes the problem faced tomorrow.
Operating effectively in those conditions requires something more precise than judgement in the conventional executive sense. It requires Astute Judgement. The capability to make the wisest available decision in full knowledge that no right answer exists, that the problem will shift the moment action is taken, and that the quality of the next decision depends on how accurately the last one was read.
Political Dexterity without Astute Judgement produces movement without direction. Astute Judgement without Political Dexterity produces clarity without consequence. Together they describe the capability profile that navigating the Galileo Dilemma at scale actually demands.
These are not qualities that can be absorbed from a framework or extracted from a training programme. The senior professionals who have developed them will confirm this readily. Most will struggle to explain precisely how. What they share is sustained exposure to consequential decisions inside complex human systems, combined with the kind of deliberate reflection that converts experience into capability rather than simply accumulating it.
That development path has always existed. What has changed is the urgency. The MIT data does not describe a gradual transition with time for unhurried capability building. It describes a five year window in which triggers of catastrophic potential will begin hitting systems that are, for the most part, not ready to navigate them.
See: Political Dexterity: The Capability That Decides What Survives and Astute Judgement: Avoiding the Trap of the Right Answer
The question this leaves
The MIT study does not address what happens inside the organisations these triggers hit. That was not its purpose. Its purpose was to quantify the threat landscape with the rigour that two hundred and seventy-two international experts across three iterative rounds could produce. On that purpose it delivers with precision.
What it leaves is the harder question. Not whether the triggers are real. The evidence on that is now as robust as expert consensus can make it. Not whether they will generate system-level resistance when they arrive. Thirty years of change research answers that with uncomfortable consistency.
The question it leaves is whether the senior professionals responsible for moving these systems have developed the capability to navigate what is coming.
Many have developed parts of it. Few have developed it deliberately at the level now required. Because the capability the situation demands has never been named with sufficient precision to develop deliberately, has never been treated as the primary determinant of organisational survival that the evidence says it is, and has never been urgent enough to compete with the demands of the role it now needs to sit inside.
That has changed. The MIT data changes it.
The Galileo Dilemma is not a rare and dramatic event that occasionally visits organisations unlucky enough to encounter radical disruption. It is the operating condition that eighteen catastrophic-threshold risks arriving simultaneously will create inside every organisation they reach. The question is not whether your organisation will face it. The question is whether you will be ready to navigate it when it arrives.
Political Dexterity and Astute Judgement are not advanced capabilities for exceptional circumstances. They are the capabilities the evidence increasingly points toward.
Most executives will read this and agree with the argument. Fewer will do anything about it before the triggers arrive. That gap between recognition and preparation is where outcomes are decided.
The window is open. It will not stay open indefinitely.
See also: The Collapsing Equilibrium and The Framework: The Architecture Behind the Work
Colin Gautrey, June 2026