Measure 03 · Pressure
When one platform or supply-chain failure hits, how many separate industries go down with it? And is that reach widening over time?
Each row is a major supply-chain or platform incident; each column a sector. A cell is filled only where public reporting confirms meaningful sector-level impact — a deliberately conservative rule that under-counts rather than over-counts. Hover a filled cell for the sourcing basis.
Reading the demarcation. The line marks mid-2023, when adversarial AI tooling (WormGPT/FraudGPT) became publicly available. Incidents below the line show broader footprints on average, but this is illustrative, not causal: the widest AI-era rows (MOVEit, CrowdStrike) stem from a zero-day and a faulty update respectively — not from AI. Sparse rows (Log4j, Okta, LastPass) reflect high cascade potential with limited documented realized impact. Sources: CISA, government advisories, Mandiant / CrowdStrike / Emsisoft analyses, and major-press reporting per incident.
The eye goes to the before-and-after split, but the sturdier finding is the column that's filled every single time: every incident, no matter the year or how far it spread, ran through the information-technology-and-cloud layer. That's not a coincidence — it's the shared substrate every other industry now sits on top of, and it ties this measure straight back to the concentration trend before it.
Read the widening reach carefully. The rise leans on two unusually broad recent incidents — MOVEit and CrowdStrike — and both trace back to a zero-day flaw and a faulty update, not to any AI technique. The mid-2023 line marks when the widest cascades show up, not a cause that produced them. And the coding is deliberately conservative: a cell is filled only where public reporting confirms real impact, so sparse rows like Log4j or Okta reflect high cascade potential with limited documented realized impact — a gap the yes-or-no coding keeps visible on purpose.