[INDIA] RBI, SEBI and IRDAI · Source-code escrow and continuity obligations for critical applications[EU DORA] ICT third-party risk testing required · In force Jan 2025[PRA] SS2/21 UK · Vendor recovery evidence required[MAS] Singapore TRM · Independent vendor recoverability expected[APRA] CPS 230 Australia · Third-party continuity obligations in force[FFIEC] United States · Source-code access and software escrow addressed in third-party contracts[ENTERPRISE] Mission-critical software procurement increasingly requires continuity evidence before contract
[INDIA] RBI, SEBI and IRDAI · Source-code escrow and continuity obligations for critical applications[EU DORA] ICT third-party risk testing required · In force Jan 2025[PRA] SS2/21 UK · Vendor recovery evidence required[MAS] Singapore TRM · Independent vendor recoverability expected[APRA] CPS 230 Australia · Third-party continuity obligations in force[FFIEC] United States · Source-code access and software escrow addressed in third-party contracts[ENTERPRISE] Mission-critical software procurement increasingly requires continuity evidence before contract
FAULTY UPDATE

The update ships. 8.5 million systems crash. The vendor cannot help

On 19 July 2024, CrowdStrike pushed a faulty content configuration update to its Falcon sensor. Around 8.5 million Windows systems crashed globally. Delta Air Lines cancelled about 7,000 flights over five days, disrupted 1.3 million passengers and reported at least $500 million in impact.

What happened

CrowdStrike's Falcon platform is endpoint detection and response software installed at the kernel level on Windows systems. On 19 July 2024, a content update contained a logic error that caused affected systems to enter a boot loop.

The outage was simultaneous and global. Airlines, hospitals, banks, broadcasters and emergency services were affected within minutes. Because affected machines could not boot normally, remediation commonly required manual intervention.

Delta Air Lines suffered the most prolonged airline disruption, cancelling about 7,000 flights over five days, affecting 1.3 million passengers and reporting at least $500 million in losses and costs.

Incident at a glance

Windows systems affected
8.5M
Delta flights cancelled
~7,000
Delta reported impact
$500M+
Passengers disrupted
1.3M
Remediation
Manual intervention on affected machines
Global sectors affected
Aviation, banking, healthcare, broadcasting

The root cause

The CrowdStrike incident illustrated a category of third-party risk that traditional escrow does not solve by itself: faulty update risk. The vendor remained operational, but one bad update caused simultaneous failure across millions of systems.

The question for every institution is whether it has a verified prior release, a tested rollback target and an executable recovery procedure when critical vendor software fails after an update.

What would Proof of Recovery have changed?

  • A current Proof of Recovery for the prior release would provide a verified rollback target.
  • A signed deployment runbook would provide a systematic recovery procedure rather than an improvised response.
  • A current SBOM and verified baseline would help security teams compare affected and known-good states.
  • Physical access might still be required, but the recovery decision and procedure could be made faster.

The vendor designed to protect critical systems became the source of failure. Recovery still depended on whether each organisation had a tested way back

The regulatory consequence

The incident intensified scrutiny of ICT change-management and recovery testing. EU DORA, the UK PRA operational-resilience framework and APRA CPS 230 each reinforce the need to test recovery from critical technology failures, including rollback and third-party disruption.

Do you have a verified rollback procedure for every critical vendor in your estate?

Book a briefing