Google wins $10 million deal for 34 years of Spirit Airlines data; flight-attendant union opposes sale

In mid-August, Google won a $10 million deal to acquire 34 years of operational data from Spirit Airlines, the low-cost carrier that declared bankruptcy in the spring. The data set includes invoices, promotion details, Wi-Fi sales, employee records and crew directives, but excludes customer data, according to a Google spokesperson. The bid outpaced a competing $7.5 million proposal from Mercor, a startup that supplies data for model training, and now awaits judicial approval.

A few days after the announcement, the Association of Flight Attendants (AFA)—which represents 5,500 former Spirit flight attendants out of its 55,000 members—filed formal opposition. Lawyers for the union argue that the package contains a massive amount of sensitive employee information and that Google’s pledged safeguards will not prevent privacy harms to tens of thousands of flight attendants who never expected their data to be used for AI training. AFA president Sara Nelson called the move “scandalous” and said the data “has no business being sold.”

Legal scholars note this is the first public clash between labor unions and corporations over the sale of employee data for AI training. U.S. law provides consumer-data protections even in bankruptcy, but comparable protections for employee data are virtually absent. “There is no line between information an employee generates and their personal information. The law didn’t stitch the gap,” said Prof. Seema Patel of the University of California, San Francisco School of Law, a researcher on technology and work. California is one of the few states with an explicit employee-data protection statute; meanwhile, “companies are celebrating it,” Patel added.

Court filings list the contents of the proposed sale as more than a million attendance-card records, over 175 thousand employee files, nearly 150 thousand tax forms, employment contracts and litigation files, 80 thousand email accounts, 17 million private OneDrive items, 20.6 million shared SharePoint files and 500 million Teams records. The proposed process requires Google to select or approve a third party to strip consumer-identifying elements from the data, but it is unclear how employee-identifying information will be handled.

The hearing on the matter has been postponed to 9 September. One former flight attendant described the potential sale as “disturbing”: “I knew they would sell any piece of consumer data. I never thought they’d be so bold to sell our private data.” Startups that specialize in selling data from bankrupt companies—such as old Slack messages, GitHub repositories and corporate drives—report revenues in the millions, reflecting a growing “data-centric” collection of human-performed tasks to train machines that may replace them.