Spirit Airlines' Data Fire Sale: Employees Left in the Cold
AI-generated image · US National Wire
As bankrupt Spirit offloads decades of internal records to Google for AI training, a legal battle reveals a disturbing gap in worker privacy protections.
Q: What is the current controversy surrounding Spirit Airlines and Google?
A: Following its bankruptcy, Spirit Airlines is attempting to sell 34 years of company data to Google. According to reporting from Wired, Google won a $10 million bid for a massive dataset that includes flight operations, invoices, Wi-Fi sales, employee records, and crew pairings. A Google spokesperson told Wired the data is intended to help improve AI models and products, though the company maintains that no personal customer information will be included in the sale.
Q: Why are former Spirit employees concerned about this transaction?
A: The Association of Flight Attendants (AFA), representing 5,500 former flight attendants, has filed a legal objection to the sale. As reported by Ars Technica, the union argues that while Google has committed to stripping personally identifying information (PII) to satisfy consumer protection laws, those laws do not adequately cover worker confidentiality. Sara Nelson, president of the AFA, told Wired that the sale of employee data is "outrageous" and has "no business being sold."
Q: What specific employee data is included in the sale?
A: According to court filings cited by Wired, the dataset is enormous. It includes: * Over 175,000 employee records and more than 1 million time-card records. * Employment contracts, litigation files, and nearly 150,000 tax forms for employees. * 80,000 email accounts and 500 million Microsoft Teams records. * 17 million individually owned Microsoft OneDrive items and 20.6 million shared SharePoint files.
Ars Technica further notes the data includes payroll information, HR records, and metrics on employee productivity and behavior.
Q: What are the risks associated with "de-identifying" this data?
A: The AFA argues that removing a name does not make sensitive information confidential. In a court filing reported by Ars Technica, the union noted that disciplinary correspondence, accommodation requests, and scheduling grievances remain sensitive even if names are stripped. Furthermore, lawyers for the AFA told Wired that AI makes it easier to relink private information across different datasets, meaning "de-identification" is not a guarantee of privacy.
Q: Who else was involved in the bidding process?
A: According to Ars Technica, the primary competitor was Mercor Corporation, an AI data and training company. Mercor submitted a $7.5 million bid and repeatedly offered to scrub the data themselves, a proposal that was rejected. Google ultimately won by offering $10 million and agreeing to pay for a court-appointed ombudsman and a third-party service to scrub the data.
Q: What is the broader implication for worker privacy?
A: Legal experts suggest this is a landmark clash. Seema Patel, a law professor at the University of California, College of the Law, San Francisco, told Wired that the law has not caught up to the reality that there is no clear boundary between an employee's work product and their personal information. Adam Schwartz, a privacy litigation director at the Electronic Frontier Foundation, told Ars Technica that the EFF opposes using personal data for new purposes, such as AI training, without explicit consent.