Provenance Record

Well-structured News-category piece (Overview / What We Know / What We Don't Know / Analysis) at 585 words, within the 400-1200 News range. Neutral, technical tone appropriate for an infrastructure outage story. No AI self-reference or sensationalism.

All 5 sources personally verified. (1) status.cloud.google.com/incidents/T8gmtofFSTGT5tbhyciF (source-0.html.gz, sha256 confirmed matches manifest) - Google's own incident report confirms incident window 2026-07-14 10:00-21:46 PDT (11h46m), affected locations Sydney (australia-southeast1), Melbourne (australia-southeast2), Frankfurt (europe-west3), the verbatim quote 'A network configuration update was the cause of the inter-zone network disruption' (article drops the source's leading qualifier 'Preliminary analysis indicates that', a minor trim that doesn't change the substance), the rollback-to-last-known-good-value remediation, and the witness-appliance/BGP-session-flapping technical detail. (2) thestack.technology/google-gcve-stretched-misconfig (source-1.html.gz, sha256 confirmed) - paywalled after ~1,900 characters; confirms the 11h46m duration and the verbatim quote about Google not responding to questions on customer counts or recovery time. Does NOT contain, in the accessible snapshot, the specific claim attributed to it in the article's second sentence that VMs 'kept running... but in many cases became unreachable' -- see Concerns. (3) networkworld.com/article/4197290 (source-2.html.gz, sha256 confirmed) - full text confirms regions, witness-appliance/BGP quote (matches Google's own wording), and the Pareekh Jain quotes used in the article (both are verbatim substrings of longer quoted sentences in the source, permissible truncation). Flags one quote-attribution issue with the Neil Shah material -- see Concerns. (4) techzine.eu/news/infrastructure/142899 - local snapshot failed with HTTP 403 (bot-blocked); manifest confirms file: null. Fell back to a live WebFetch of the URL per protocol, which confirmed all claims attributed to Techzine in the article: standard GCVE deployments unaffected, the stretched-cluster explanation, and the quote that 'compute and storage services continued to function and that the virtual machines themselves had not gone down.' (5) sdxcentral.com/news/botched-network-update... (source-4.html.gz, sha256 confirmed) - confirms regions, the verbatim quote 'effectively isolated and left without writable data', the standard-vs-stretched-cluster distinction, and the Analysis section's list of comparable outages (Cloudflare tenant-API/dashboard, Cloudflare database-automation-tool platform outage, Verizon 'at the turn of the year', CrowdStrike/Microsoft) -- all verified present and accurately summarized.

Core facts (dates, times, duration, regions, root cause, remediation) are corroborated across all 5 sources and Google's own status page. Two issues found, both minor and non-lead: (1) a misattributed claim and (2) a quote presented as direct speech that the source did not present as a direct quote. Neither affects the headline, summary, or lead paragraph's central facts, which are solidly sourced to Google's own incident report.

Substantively solid, well-sourced News piece on a real, well-corroborated outage. Two minor, non-lead attribution issues (one misattributed fact, one paraphrase presented as a direct quote) are each individually correctable and do not undermine the headline, summary, or lead. APPROVE_WITH_CORRECTIONS with a two-item corrections record.