Google’s winning bid for a large Spirit Airlines dataset has opened a sharp privacy fight over what happens to employee records when a bankrupt company sells data. The deal is built around stripping personally identifying information before Google receives the files, but former flight attendants argue that removing names is not the same as protecting confidential workplace information.
The dispute centers on a simple tension: the privacy safeguards described in the sale are aimed at PII, while much of the material being sold is employee-facing. For workers, the concern is not only whether a record names someone. It is whether sensitive employment history, internal communications, payroll details, or workplace behavior data can still reveal information that should not travel into another company’s AI and product systems.
What Google bought in the Spirit Airlines auction
Last Friday, Google won an auction to acquire a large amount of Spirit Airlines data. The dataset does not include personal information or customer data, according to the source article, but it does cover a broad sweep of the airline’s employment and workplace record.
The sale followed Spirit Airlines’ bankruptcy and the decision to auction off a massive dataset to the highest bidder. At the auction, Spirit debtors appeared to favor bids that would avoid frustrating customers affected when the airline abruptly shuttered on May 2.
The auction was held on August 14. Google opened with a $5 million bid and said from the beginning that it would pay for a third party to scrub the data. Other bids that sought additional consumer data, including a customer list Spirit would not sell, were rejected. Bids seeking PII were also set aside after the first round.
The dataset included Spirit computer programs, applications, and code. It also included worker data spanning decades, including approximately 100 million employee emails, HR information, payroll data, and data measuring employee behaviors, activity, and productivity.
Google ultimately won with the highest price, $10 million, and by agreeing to cover the cost of a third-party service to scrub the data so the sale would comply with consumer privacy laws. A backup bid with similar terms for $7.5 billion was accepted from Mercor, should Google fail to complete the purchase.
Why flight attendants objected
The Association of Flight Attendants (AFA), a collective bargaining unit representing Spirit workers, filed a limited objection Tuesday. The objection does not seek to disrupt the sale. Instead, it argues that Spirit debtors protected consumers more clearly than workers.
AFA’s core point is that the deal’s privacy design is based on consumer protection laws, even though the dataset is largely about employees. The union argues that those laws do not address worker confidentiality, creating what it describes as a major gap in the transaction.
“The privacy architecture of this transaction is consumer-facing; its payload is disproportionately employee-facing,” the AFA argued. “Hence, the employee data is far more confidential than the customer data, yet receives far less protection than the customer data.”
The objection says the Sale Agreement does not require anyone to screen for, separate, or restrict use of confidential employee information that workers consider sensitive. AFA drew a distinction between de-identification and confidentiality: removing a name may make a record harder to trace, but it does not change the nature of what the record contains.
Examples in the filing include disciplinary correspondence, crew training deficiency, leave or accommodation request, internal Teams exchange about staffing or scheduling grievances, and payroll adjustment history. AFA’s argument is that each remains sensitive employment information even if the worker’s name has been removed.
The limits of removing PII
Google agreed to use a court-appointed ombudsman to oversee a process that strips personally identifying information from the data before transfer. Google also agreed to keep the data in de-identified form and never intentionally re-identify it. If Google later sells access to the data, third parties would supposedly be bound by the same terms.
Those commitments are important, but they do not resolve the flight attendants’ concern. Their worry is that confidential worker information may remain visible inside the dataset even after direct identifiers are removed. A record can be anonymous in one sense and still expose sensitive patterns, claims, disputes, performance issues, or workplace communications.
The AFA also raised concern that Google could combine the worker data with other Google datasets and re-associate workers with their Spirit records. The union does not claim that any particular record can be re-identified. Its position is that the risk is not speculative, especially when scrubbed data can be compared with publicly available information.
The source article notes a Georgia Law researcher’s 2017 point that increasingly powerful computer hardware has made it easier to combine public data with scrubbed data to de-anonymize it. If the Spirit dataset is used for AI training, AFA argues, then it will inevitably be combined with other data.
Google’s position on the deal
Google says it is not seeking to identify people connected to Spirit Airlines. A Google spokesperson told Ars that the company acquired part of an enterprise dataset from Spirit Airlines because it can help improve products and AI models.
“We acquired part of an enterprise dataset from Spirit Airlines, which can be helpful in improving our products and AI models,” the spokesperson said. “We will not receive any personal information from this dataset. Any data we receive will be rigorously scrubbed of any personally identifiable information by a third party before receipt.”
That statement addresses PII directly. The flight attendants’ objection, however, focuses on information whose sensitivity may not depend on names. AFA argues that small groups, crew bases, training outcomes, investigations, compensation adjustments, and internal discussions could still be exposed in ways workers would not expect.
The union also acknowledged that Google’s public commitments against re-association are meaningful and “were not obviously required.” But AFA’s broader point is that a pseudonymized dataset can still disclose sensitive workplace information. In its view, the consequence is that information whose sensitivity has nothing to do with names would pass through untouched.
Why the case matters for workplace data
The Spirit Airlines data sale shows how bankruptcy can turn years of workplace systems into assets for auction. It also highlights a privacy problem that becomes more visible when enterprise data is purchased for AI and product development: employees may not have agreed to have old emails, HR records, payroll history, or productivity data used for a new purpose.
Adam Schwartz, privacy litigation director for the Electronic Frontier Foundation, told Ars that the sale alarmed privacy advocates. He said EFF opposes using a person’s data for a new purpose without first getting consent, which he said does not happen when a bankrupt company sells employee emails to become AI training data.
The immediate fight is about Spirit Airlines, Google, and former flight attendants. The broader issue is how companies and courts define privacy when the data is no longer plainly identifiable but still deeply revealing. For workers, the question is not only whether Google receives names. It is whether de-identified workplace records can still carry the substance of private employment lives into systems built for entirely different ends.