ong-ot-exposure-report-2026

Limitations

Numbered for reference from the report and the verification checklist. Written before the discussion section.

  1. Disclosed, not deployed. Counts measure what vendors and CISA disclose. They do not measure the installed base in US ONG facilities, or patch status. Vendors with mature PSIRTs publish more advisories.
  2. Scope is a proxy. CISA’s Energy tag covers electricity as well as oil and natural gas. Some electricity-only products remain in scope, and some ONG-used products tagged only with other sectors and absent from the v1.1 taxonomy are excluded. The EV-charging exclusion is a title/vendor pattern (16 advisories removed).
  3. Input double-ingestion. The published v1.1 dataset contains every CVE-advisory pair twice (27,944 rows, 13,791 unique pairs), because the ICS Advisory Project master file and its per-year files were both ingested. This report keeps master-file rows only. The QA report shows no pairs lost. Counts cited from v1.1 itself are about double.
  4. Advisory-level severity and CWE. A roll-up advisory contributes many CWEs, and all of its CVEs share one severity. Figures 2 and 6 and Table 1 report roll-up sensitivity.
  5. Point-in-time enrichment. KEV, EPSS and SSVC are as of 2026-09-12 for every year, so recent CVEs have had less time to accumulate exploitation evidence and SSVC coverage (2026 SSVC coverage 51.4% vs 2025 87.0%).
  6. Mirror dependency. Advisory fields come from the ICS Advisory Project transcription, not directly from cisa.gov. The spot checks are the control.
  7. Partial year. The edition covers 1 Jan - 10 Sep 2026. The December rebuild will change every number.
  8. KEV component attribution is a manual reading of CVE descriptions (config/kev_component_attribution.csv). All 7 were confirmed by the author against NVD on 2026-09-26.
  9. Remediation signal. Machine-readable CNA remediation text covers only a minority of CVEs. “No text” is treated as unknown.
  10. The build environment could not reach cisa.gov, first.org or nvd.nist.gov (egress policy). No source was re-fetched. The report relies entirely on the DOI-registered v1.1 snapshot, and verification against live CISA pages is the author’s step.