Escalation Lens

Analyze how cyber operations move through phases of escalation, from initial access to strategic consequences, and how states signal, restrain, or intensify.

Unpeace Spectrum

Each incident is positioned on a 0–100 unpeace axis reflecting peak escalation tier and the thresholds crossed from the closed list (governance is a separate axis). Hover or tap an incident to inspect.

Stable (0–30)
Contested (30–60)
Escalatory (60–100)
0102030405060708090100

Escalation Categories

Cyber operations are classified along a six-tier escalation ladder. Each tier describes a qualitative shift in the nature and severity of the operation, not a linear progression.

ProbingTier 1

Scanning, reconnaissance, and testing defences without meaningful disruption. Establishes access potential.

0 incidents in dataset

IntrusionTier 2

Unauthorized access gained and maintained. Data may be collected but no operational effect is imposed on the target.

11 incidents in dataset

DisruptionTier 3

Operations that temporarily deny or degrade services. Systems recover without permanent damage, but operational continuity is interrupted.

5 incidents in dataset

DegradationTier 4

Sustained impairment of target capabilities. May involve data destruction, prolonged outages, or significant economic cost.

10 incidents in dataset

DestructionTier 5

Irreversible damage to systems, data, or infrastructure. Recovery requires rebuilding rather than restoring.

5 incidents in dataset

Strategic ImpactTier 6

Operations with national-security or international significance, affecting critical infrastructure, economic stability, or interstate relations.

2 incidents in dataset

Compellence vs. Deterrence

Thomas Schelling distinguished two coercive logics: compellence (forcing an adversary to change behaviour through imposed costs) and deterrence (dissuading action by signalling capability and willingness to retaliate). Cyber operations often blur this distinction, the same intrusion can serve as intelligence collection and a latent threat. This split offers an analytical lens, not a definitive classification.

Deterrence

Operations that signal capability without imposing irreversible harm

SolarWindsMarch 2020 – December 2020

Restraint observed: Operated within traditional espionage norms, collection, not disruption.

Espionage30

Sony PicturesNovember – December 2014

Restraint observed: Targeted a single corporation, not government or critical infrastructure.

Destructive50

Exchange/HafniumJanuary – March 2021

Restraint observed: Initial phase was narrowly targeted espionage.

Espionage50

Shamoon / AramcoAugust 2012

Restraint observed: OT and production control systems were air-gapped and unaffected.

Destructive50

Bangladesh BankFebruary 2016

Restraint observed: A typo in a transfer request triggered manual review, limiting losses.

Hybrid40

Australia ParliamentJanuary – February 2019

Restraint observed: No destructive or disruptive actions taken, activity consistent with intelligence collection.

Espionage30

Iran Nuclear Cyber2020 – 2021

Restraint observed: Incidents were narrowly targeted at specific nuclear facilities.

Sabotage40

Taiwan Telecom2022 – 2023 (disclosed 2023)

Restraint observed: No disruptive or destructive actions observed.

Espionage20

MGM / Scattered SpiderSeptember 2023

Restraint observed: Financially motivated, no geopolitical or destructive intent.

Ransomware40

Storm-0558May – July 2023

Restraint observed: Activity consistent with targeted intelligence collection, not disruption.

Espionage30

Volt Typhoon2023 – 2024 (disclosed 2024)

Restraint observed: No disruptive or destructive actions observed, activity consistent with preparation, not execution.

Espionage20

Midnight BlizzardNovember 2023 – January 2024 (disclosed January 2024)

Restraint observed: Activity consistent with espionage, no destructive or disruptive payload deployed.

Espionage20

India–Pakistan Cyber2016 – 2019 (multiple incidents)

Restraint observed: Operations remained within espionage parameters, no destructive payloads documented.

Espionage20

Gaza Cybergang2018 – 2022 (ongoing, landmark incidents)

Restraint observed: Operations remained focused on intelligence collection, not disruption.

Espionage20

Belgacomcirca 2010 – disclosed September 2013

Restraint observed: Activity consistent with intelligence collection, not disruption, no destructive payload, no manipulation of communications integrity disclosed.

Espionage20

OPM2014 – disclosed June 2015

Restraint observed: Activity confined to data exfiltration; no destructive payload, no manipulation, no public release of the stolen data.

Espionage30

APT12006 – disclosed February 2013; indictment May 2014

Restraint observed: Activity confined to economic and business intelligence collection; no destructive payload, no manipulation of data.

Espionage20

Salt TyphoonAccess established earlier; disclosed September–December 2024 (note: investigation and disclosures ongoing as of May 2025)

Restraint observed: Activity to date consistent with intelligence collection; no public evidence of destructive payload or manipulation of communications integrity.

Espionage20

FlameActive circa 2007 – disclosed May 2012

Restraint observed: Activity confined to intelligence collection; no destructive payload, no manipulation of operational systems.

Espionage20

Cloud HopperActive circa 2014 – publicly disclosed April 2017; indictment December 2018

Restraint observed: Activity confined to intellectual-property and business-intelligence collection across MSP client networks; no destructive payload, no manipulation.

Espionage20

Analytical note: This classification uses the operation's primary coercive function as a heuristic. Many operations serve both logics simultaneously, an espionage campaign that pre-positions destructive capabilities deters through demonstrated access while also enabling future compellence. The distinction is most useful as a teaching tool for examining how states signal intent through cyber operations, not as a rigid taxonomy.

On reading escalation

Escalation is not a conveyor belt, incidents do not inevitably progress from probing to destruction. Restraint is as analytically important as escalation. States frequently choose not to escalate, and understanding those choices requires examining governance constraints, norm commitments, and deterrence calculations alongside technical capabilities.

The categories and positions shown here are derived from the dataset's incident records. They reflect analytical judgement, not predictive modelling. All assessments should be read alongside the source material linked in each case.