A “Credible Warning” Must Trigger Deterrence: The Global Compact and AI Sentinel That Could Prevent the Next Ukraine

In the winter before the full-scale invasion, the evidence was already piling up—satellite images of Russian units, the telltale logistics of a force preparing not for exercise but for occupation, and, according to reporting later captured in The Guardian’s phrase “a war foretold,” unusually specific US and UK intelligence that suggested Vladimir Putin’s decision had hardened into a plan. Yet across capitals and living rooms, the prevailing instinct was to shrug. It was bluff. It was leverage. It was theater.

Then, on 24 February 2022, the theater became artillery. For Ukrainians, the consequences were immediate and intimate: the rush to train stations, the last phone calls before sieges, the new geography of exile. For Europe, it was the largest land war in generations, a shock that ricocheted through energy prices, food markets, defense budgets, and political cohesion. And for the world, it exposed a brutal paradox of modern security: even when democracies are warned, they often lack a disciplined way to turn warning into action fast enough to change an aggressor’s calculation.

The global problem is not only intelligence collection. It is the “belief gap”—the space between credible warning and credible response.

The belief gap: why “we saw it coming” wasn’t enough

Part of the failure was human. Leaders are paid to project calm; admitting an invasion is likely can spook markets, trigger capital flight, and create panic that harms a country even if war is averted. In Kyiv in late 2021, skepticism was not simply denial; it was also economic triage. In European capitals, caution was shaped by bitter memory—intelligence that had once been oversold, wars justified with certainty that later looked like spin. In Washington and London, even unusually explicit pre-invasion warnings collided with a basic democratic constraint: officials could not show the public their best sources without burning them.

But part of the failure was structural. Intelligence agencies are built to assess. Governments are built to deliberate. Alliances are built to negotiate consensus. An aggressor needs one decision; a coalition needs many—across parliaments, procurement systems, legal authorities, and public opinion. By the time a skeptical public is convinced, the window for prevention is already closing.

That is why the most damning postwar sentence is not “we didn’t know.” It is “we knew, but we didn’t move.”

A workable solution: a Deterrence-on-Warning Compact, backed by an AI Sentinel

What would it look like to design the missing machinery—an automatic bridge from credible warning to rapid deterrence—without sliding into secrecy or overreaction?

Start with a standing Deterrence-on-Warning Compact among willing democracies and partners: a pre-negotiated, legally and financially prepared package of actions that activates when warning thresholds are met. The key is not bravado—no blank checks, no automatic war. It is speed, reversibility, and predictability: actions that make an invasion harder to launch and harder to win quickly, while keeping diplomatic off-ramps open.

Then pair it with a second innovation that addresses the trust problem head-on: an AI-enabled “sentinel” platform that fuses open-source intelligence—commercial satellite imagery, logistics signals, cyber telemetry, social media indicators, shipping and financial data—with classified assessments in a way that can be partially demonstrated to the public. The point is not to outsource statecraft to algorithms. The point is to make warning legible enough that leaders cannot hide behind ambiguity—and publics cannot be asked to accept faith without evidence.

It is easy to hear “AI” and imagine science fiction. But the underlying idea is familiar in other domains: detect rare, high-stakes signals in oceans of noise. High-energy physics collaborations do this routinely, sifting petabytes of detector data to isolate vanishingly rare events. Geopolitics is messier than particle collisions, but the need is analogous: a system that continuously integrates weak indicators into a transparent probability forecast, updated as new evidence arrives and audited after the fact.

How it would unfold: from the first warning to a changed Kremlin calculus

Picture the next crisis—somewhere on NATO’s periphery, or a partner facing coercion. The sentinel platform begins to shift: commercial satellites show the staging areas filling; rail traffic and fuel depots thicken; cyber probes intensify against ministries and power grids. Classified reporting—kept protected—pushes the confidence higher still. Instead of a familiar fog of debate, the Compact’s trigger is met.

Within seventy-two hours, not weeks, the first deterrence tranche moves.

Ukraine in late 2021 needed, above all, time and survivability in the first month: more layered air defense, more counter-drone systems, more secure command-and-control, more anti-armor munitions, and training pipelines that could expand forces quickly without collapsing the state’s finances. A Compact makes that kind of surge procedural, like disaster relief. Defense logistics are pre-contracted. Stockpiles are earmarked. Training teams and maintenance packages are already funded. Cyber defense assistance is not an afterthought—it arrives early, visibly, and at scale, raising the cost of sabotage and confusion operations that typically precede armor.

At the same time, the Compact stabilizes the target’s economy, because disbelief is often a form of financial self-defense. A pre-arranged G7-style insurance backstop—designed in peacetime—helps prevent a run on currency and bonds, keeps public-sector salaries paid, and supports mobilization without panic. The message to citizens becomes steadier: preparedness is what keeps life normal, not what ends it.

Then comes the deterrence lever that has repeatedly failed when used only as a threat: sanctions. The Compact pre-defines reversible “circuit breakers” tied to observable triggers—cross-border movement of major formations, dispersal of missile units, a large cyberattack on critical infrastructure. The goal is to remove the fatal delay between “if you invade, we will…” and the reality that sanctions packages often take weeks of political wrangling. If the aggressor steps back, measures pause. If the aggressor escalates, the cost curve rises predictably and immediately.

And crucially, the information strategy changes. Some intelligence will always remain classified. But the sentinel’s fused, open-source layer can be shown: timelines, confidence bands, imagery, and the reasoning chain. The public sees the pattern, not just the pronouncement. That makes it harder for an aggressor to flood the zone with denial—and harder for democracies to talk themselves out of what their own eyes can verify.

What success looks like by 2030: fewer “surprise” wars, more credible off-ramps

No compact can guarantee that an autocrat will not choose war. But deterrence is not a magic wall; it is a staircase of obstacles. The measurable aim is to defeat the “quick win” theory—decapitation strikes, rapid seizure of a capital, paralysis through cyber and shock. If a would-be invader believes the first thirty days will be slow, costly, and humiliating, the decision calculus changes.

Success would be visible in timelines as much as in headlines. By 2027, Compact members could run semiannual “warning-to-action” exercises the way cities rehearse earthquake response: logistics, finance, cyber, communications, parliamentary briefings, and after-action audits. By 2030, the sentinel platform could make coercive buildups harder to disguise, while the Compact makes them harder to exploit. The price tag—several billion dollars a year for a multinational system—would be trivial next to the hundreds of billions democracies have already spent responding after the fact, not to mention the human costs borne by the attacked.

Just as important, trust would become enforceable. A post-crisis, cross-party audit mechanism—built into the Compact—would review declassifications, thresholds, and actions, guarding against politicized intelligence. Democracies do not need less skepticism; they need institutions that can withstand skepticism and still act.

The call: build the bridge before the next warning arrives

The lesson of a “war foretold” is not that intelligence is pointless. It is that intelligence without a rapid, credible conversion into deterrence is merely prophecy. We should not accept a world in which the best our governments can offer threatened nations is: “We saw it coming.”

Parliaments should demand a Deterrence-on-Warning Compact that is funded, rehearsed, and auditable. Governments should invest in a sentinel capability that fuses open-source and classified insight into warnings the public can actually weigh. And citizens should insist—politely, relentlessly—that when credible alarms sound, leaders move first and argue while moving, not the other way around.

History will not judge us by what we knew in private. It will judge whether we built systems capable of acting on what we knew—fast enough to keep the forecast from becoming the funeral.