Spirit's estate priced its employee records at 1.6 cents an item. A court wants a closer look.

A federal bankruptcy court delayed approval of Google's $10 million purchase of Spirit Airlines' digital records after former flight attendants challenged the privacy terms, according to The Wall Street Journal ^[1]. The hearing moved from August 19th to September 9th after the Association of Flight Attendants-CWA filed its objection, Reuters reported ^[2], giving the court three extra weeks ^[3] to decide whether removing identifiers is enough to turn an employer's internal history into AI training inventory.

The inventory is large and it is mostly workplace history. Spirit proposes to transfer roughly 100 million emails across 80,000 accounts, 500 million Microsoft Teams items, 17.1 million OneDrive items and 20.6 million SharePoint items ^[4]. That is about 638 million messages and documents ^[5], or roughly 1.6 cents per item at the agreed price ^[6]. The schedule also lists 175,658 employee records dating to August 1986, 3.4 million payroll records, 148,018 employee tax forms, crew training data and more than 5 million crew pairings ^[7] - about 19 payroll records for every employee record ^[8]. Google would also receive 516 source-code repositories holding approximately 30 million lines of code, plus commit histories, bug reports and development discussions ^[9], an average of roughly 58,000 lines per repository ^[10]. Operational and commercial records cover aircraft logistics, pricing, revenue management, audits, fraud, project management and customer-service workflows ^[11]. Google says the material could improve its products and AI models ^[12].

Spirit drew the exclusion line at customers, keeping 97.5 million passenger profiles and 50.2 million Free Spirit loyalty records out of the deal ^[13], about 148 million records withheld ^[14] while the employment file goes across.

The technical crux is a contract term. Google told Axios that a third party would rigorously scrub personally identifiable information before delivery ^[15], and the agreement requires Google to keep the data de-identified and not intentionally associate it with a person or household ^[16]. But the same agreement requires the de-identification process to maintain referential integrity, so records from different systems stay linked ^[17]. That linkage is what makes the dataset worth buying: a model can trace how an operational event moved through scheduling, communications, payroll and management systems ^[18]. It is also the union's objection. AFA-CWA argues that stripping direct identifiers does not remove details about discipline, medical accommodations, performance, union activity or grievances ^[19], and that a small crew group can stay recognizable from its base, assignments, training history and conversations ^[20].

The enforcement terms are thinner than the technical ones. The agreement bars only intentional re-identification ^[21], permits onward transfer to third parties that accept the same restriction ^[21], and gives flight attendants no third-party beneficiary rights, according to the union's filing ^[22]. AFA-CWA wants all flight attendant information removed, with a fallback demand for separate review of personnel, payroll, medical, leave, disciplinary, performance and union records, plus a ban on using the data to profile or score individuals or identifiable groups ^[23].

Privacy was already a line item in the bidding. Google opened at $5 million and agreed to pay for the third-party de-identification process ^[24], ending at double its opening number ^[25], against Mercor.io as backup bidder at $7.5 million ^[26].

Watch the protocol, not the ruling. As of an August 18th update, AFA-CWA said the data had not been delivered and the de-identification method had not yet been designed ^[27]. A court approving a scrub that nobody has specified is approving a promise, and referential integrity is the term that decides what the promise is worth.