← Breach analysis by the Dual Vulnerabilities framework

US National Weather Service weather.gov Hack and Data Leak (Kosova Hacker's Security, 2012)

2012 · United States · O - Public administration and defence · Website intrusion via web-application vulnerability and data leak

Analysed 2026-11-07 by Hermes

Open this case in the tool →

What happened

Website intrusion via web-application vulnerability and data leak

Dual vulnerability type

Expertise to Execution

Evidence coverage: 1/6 stages, weight 8/60

Decisive stage: entry (clearest evidence) · Alternative: none defensible

Why this classification

(1) weather.gov is the public web estate of the US National Weather Service (NOAA, Department of Commerce), whose mission is warnings for the protection of life and property; the expectation at a federal agency of this kind is that its internet-facing web applications are built and kept free of well-known web vulnerability classes - sanitised inputs, validated file includes, encoded output - before and while serving public traffic. (2) In practice the previously unknown, self-attributing group Kosova Hacker's Security exploited a local file inclusion vulnerability on weather.gov to read files off the server (/etc/passwd, system and network configuration files, administrative account names able to log in via console) and published the material on Pastebin on 16 October 2012; administrators fixed the vulnerability only around the leak, and a cross-site scripting hole was subsequently discovered on the same site, showing a second coding defect still open; NOAA made no public statement, AFP could not obtain comment, and the intrusion became known entirely through the attacker's own dump. (3) The unmet expectation sat in the material column: the remedies for LFI and XSS are routine specialist knowledge, not policy choices, the case file records no zero-day, and the two-day fix shows how readily the remediation could have been applied - yet the secure practice was simply not realised in the production web application, and no source shows a doctrinal artefact that failed to translate, a leadership decision deferred, or an expert warning that leadership ignored. That is a knowing-doing gap within the material column, Expertise to Execution, and it rests on the entry stage, the only stage with a documented victim-side gap; the entry vector here is documented and load-bearing. The later stages do not carry the classification: containment stayed on the web server's own files, detection is undocumented (patched within two days, origin of awareness unknown), recovery was quick with disruption under 24 hours, and disclosure was attacker-driven with the victim silent - a condition no marker's exclusion clause lets the evidence code as a notification failure. RM29 (Expertise to Execution lens): the weather.gov application was built without basic input sanitisation, and the LFI that enabled the intrusion plus the XSS hole found on the same site right after show that secure-coding practice known to developers was not applied to this production federal site (causal role: Enabling). RM18 (Expertise to Execution lens): a non-zero-day vulnerability remained exploitable on a public federal website until the attacker's leak forced its removal, and the two-day fix shows the readily available remediation had simply gone unexecuted, so exposure persisted until external pressure ended it (causal role: Enabling).

Evidence by stage

Risk markers

Sources

Notes

EuRepoC row; victim, country and year from receiver_country (United States) and start_date (01.10.2012). Date discrepancy: the dataset start date is 1 October 2012, while contemporary reporting (The Hacker News, 16-10-12, describing the hack as done two days earlier) places the intrusion in mid-October 2012; the year is unaffected. Attribution rests solely on the group's self-attribution via its Pastebin dump and message; security firms (Acunetix, Sophos, quoted by AFP) inspected the leaked data but Sophos noted little was known about the group; the dataset's initiator_country United Kingdom is not corroborated by any reporting opened for this row. The dataset codes the incident as Disruption with data_theft none, while reporting describes a leak of system/configuration files, directory structures, partial login credentials and admin account names (no personal or scientific data reported). No NOAA public statement about the intrusion was found in the 2012 record; AFP reported the weather service did not respond to a request for comment. Not to be confused with a separate, later NOAA/NWS intrusion of late September 2014 reported by the Washington Post as China-linked - a different incident, not used as evidence here. The original Sophos Naked Security article of 19-10-2012 no longer opens at its 2012 URL (it redirects to the current Sophos blog index), so its findings are cited as reported by SecurityWeek (AFP) and The Register; that Sophos URL is therefore not cited.