← Breach analysis by the Dual Vulnerabilities framework

City of Philadelphia email account breach (Unattributed, 2023)

2023 · United States · O - Public administration and defence · Data exposure via compromised email accounts

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

Data exposure via compromised email accounts

Dual vulnerability type

Strategy to Leadership

Evidence coverage: 3/6 stages, weight 19/60

Decisive stage: response (clearest evidence) · Alternative: Expertise to Execution

Why this classification

(1) The City of Philadelphia is a municipal HIPAA covered entity that handles sensitive personal and health records and maintains a HIPAA & Privacy Law Unit; the strategic expectation it is held to - and states itself in its notices - is to safeguard information in its care and to notify affected individuals promptly. (2) In practice, an unattributed actor held access to City administrative email accounts from 26 May to 28 July 2023, including after the City opened an investigation into suspicious email activity on 24 May 2023; the City identified PHI in the accounts on 22 August 2023, posted only a website notice on 20 October 2023, and mailed written notices to the eventual 35,881 affected individuals on 16 May 2024 (PHI) and 8 July 2024 (personal information) - more than a year after awareness - while asserting it 'immediately took steps' to secure its systems. The entry vector was never explained and is not load-bearing. (3) This is a doctrinal-column gap: the expectation (prompt notification, protection of sensitive records) exists at the strategic/legal level and the City's own decision-makers administer it, yet the timing decision - notify individuals only after a programmatic and manual review, result validation and address location were finished - left the population unnotified for about thirteen months, and the immediate-action narrative diverges from the City's own documented discovery (28 July 2023 per the Maine AG filing) and notification dates. Intent and enacted behaviour diverge within leadership, which is Strategy to Leadership. The dwell time could be read as an Expertise to Execution detection gap, but no source documents the eradication or alert-handling decisions behind it, so it does not win the decisive-stage choice. (4) RM39 (Strategy to Leadership lens): leadership disclosure behaviour diverged from the legal expectation it administers - awareness 24 May 2023 and written notice to 35,881 individuals only in May/July 2024, an unexplained delay the press highlighted (causal role: Constraining). RM21 (Strategy to Leadership lens): the City's email environment had already been compromised twice in 2020 - a phishing attack with access to city inboxes that led to notice to 49,000 residents, and the DBHIDS-related phishing breach affecting over 108,000 people with mailbox access spanning March to November 2020 - and a comparable mailbox breach recurred in 2023, showing attention and accountability did not persist after the prior incidents (causal role: Perpetuating). RM38 (Strategy to Leadership lens): unauthorized access persisted from 26 May to 28 July 2023 although an investigation was already open from 24 May, indicating full eradication was not prioritised to completion during the response; the leadership mechanism itself is only inferable (causal role: Amplifying). RM16 (Strategy to Leadership lens): privacy commitments were not reflected in data practice - diagnoses, claims data and SSNs were reachable in routine administrative mailboxes, and the City needed a manual sweep running into 2024 to learn which records related to whom (causal role: Amplifying).

Evidence by stage

Risk markers

Sources

Notes

Entry vector never disclosed: city officials 'have yet to explain how the attackers breached the City's email accounts and why they delayed the disclosure for five months' (BleepingComputer); classification rests on the disclosure/response stage. Timeline ambiguity is in the City's own documents: the 2023 notice says it became aware of suspicious activity on 24 May 2023, while its Maine AG filing records Date Breach Discovered as 07-28-2023; The Record flagged the unexplained date discrepancy between awareness (24 May) and the access window starting 26 May. The case file's affected_entities 1-10 counts affected organisations, not individuals - the individual count is 35,881 (Maine AG filing, 15 Maine residents). The eurepoc_extra.intelligence_impact value is dataset boilerplate (mentions data corruption) that conflicts with the access/exfiltration narrative; has_disruption False and functional_impact (no system interference) match the sources. The unattributed actor does not lower confidence (DV assessed victim-side). The two Checkpoint threat-report links in source_url were not opened and are not cited. Link check: maine.gov blocks automated access (HTTP 403); the page was opened and read during research. No regulator investigation or fine on this incident was found during research.