- Google’s $10 million bid topped a $7.5 million offer from AI data company Mercor, but the transaction remains subject to a bankruptcy-court hearing scheduled for August 19, 2026. [1] - The package includes roughly 100 million emails and 500 million Microsoft Teams messages, along with financial, operational, maintenance and software-development data. [1][2] - Passenger profiles, loyalty accounts, credit-card-member data, customer-service calls and chat records are excluded from the asset schedule. [2] - The agreement requires de-identification before delivery and bars Google from intentionally associating the data with people or households. It does not state a fixed retention period or require individual employee or customer notice.
[2] Google has won a $10 million bankruptcy auction for Spirit Airlines’ de-identified internal business data, including roughly 100 million emails, 500 million Microsoft Teams messages, airline operating records and software. The auction result was disclosed in a notice filed August 14 with the U.S. Bankruptcy Court for the Southern District of New York, but the transaction cannot close until the court approves it at a hearing scheduled for August 19, 2026. [1] ()
Google said it plans to use the enterprise dataset to improve its products and AI models. Under the proposed sale agreement, a third-party de-identification agent must process the data before delivery, and Google publicly commits not to intentionally associate the resulting dataset with any person or household. [1] [2] ()
What Google is buying #
The court filing describes a package broader than email and chat logs. The asset schedule includes financial models, invoices, vendor information, corporate-development files, audit materials, presentations, airline operations systems, maintenance records, fuel data, crew pairings and disruption-management records. [2]
The filing lists about 100 million emails and 500 million Teams messages, as well as documents and data held in collaboration and storage systems. It also includes software repositories and related development records. [1][2]
The package includes airline business-system data covering areas such as flight operations, maintenance, fuel and financial accounting. The filing distinguishes those operational records from passenger profiles and other customer datasets that are not part of Google’s purchase. [2]
The customer-data boundary #
The asset schedule excludes passenger profiles, Free Spirit and Saver$ Club membership information, earned miles, credit-card-holder information, active customer email addresses and customer marketing activity. It also excludes customer-service chat sessions, call recordings, phone numbers and website analytics. [2]
The agreement separately says the assets cannot include information associated with a consumer or other personal data under applicable data-protection laws. Before transfer, a de-identification agent must transform the data so it cannot reasonably be linked to a particular consumer, while preserving referential integrity across the dataset. [2]
That language leaves an important implementation question: the schedule contains categories of internal employment and business records, but the agreement requires any information meeting its definition of personal data to be excluded or transformed before delivery. The public filing does not identify the de-identification vendor or provide a sample of the processed data. [2]
What the agreement restricts #
Google commits to maintain and use the data in de-identified form and not intentionally associate it with any person or household. It may transfer the de-identified data to third parties if those recipients are contractually bound by the same restrictions. [2]
The agreement does not impose a stated expiration date or retention schedule. It also does not limit use to AI training alone: Google’s public description refers more broadly to product development and AI-model improvement. The filing does not describe individual notice to former employees or customers. Its notice provisions concern communications among Spirit, Google and their representatives, while the bankruptcy process provides notice to parties in the case. [1][2]
NIST guidance describes de-identification as a process for reducing disclosure risk, not an absolute guarantee that data can never be linked to individuals. It recommends evaluating re-identification risk, including risks created by combining quasi-identifiers across datasets. [3] That consideration is relevant to a package combining communications, operational histories, employment-related records and detailed business transactions, even with customer profiles excluded.
Mercor, an AI-powered hiring company, submitted the next-highest bid at $7.5 million and is designated as the alternate buyer if Google does not complete the transaction. The bankruptcy court’s hearing is scheduled for 11 a.m. Eastern time on August 19. [1][2]
Companies mentioned #
Further sources #
The stories that matter, in one email. Free — unsubscribe anytime.