Spirit’s $10M Data Move Sparks Backlash

Photo: Linda Parton / Shutterstock

Union leaders and senators say Spirit Airlines planned a $10 million transfer of vast employee records to Google for AI training, putting worker privacy on the line.

Story Snapshot

Spirit’s proposed deal with Google would include millions of employee communications and records, unions say.
Workday and other vendors describe privacy controls and de-identified data, but independent proof of safety is limited in the record.
Experts warn most workers never agreed to sell their messages or emails for AI training.
Lawmakers urged a halt until protections prevent identification, profiling, or misuse of worker data.

What Sparked The Fight

Senator Maria Cantwell and colleagues said Spirit planned to transfer employee-related records to Google for artificial intelligence training and urged strong safeguards first. The union representing flight attendants said workers did not consent to any sale of their data. The union also warned the transfer could include about 100 million emails and 500 million Microsoft Teams messages, plus payroll, tax, timecard, and contract data. Spirit and Google have not publicly released the full terms cited by lawmakers.

Labor groups said the transfer would turn private workplace records into training fuel for a technology company, raising risks for profiling and error. Privacy advocates said even “de-identified” data can allow inferences about individuals or small teams, which can still cause harm. Lawmakers pressed Spirit and Google to ensure that disciplinary records, medical or accommodation requests, and pay information receive strong protection against exposure and misuse.

What Companies Say About “Anonymized” Use

Workday, a major workplace-data platform, says it and authorized partners may use de-identification, pseudonymization, and aggregation for research, analytics, and product development. Workday also says customers keep data ownership, can apply privacy controls, and can trigger deletion at the end of a contract. Workday describes a review process that removes data judged not needed for features, and claims these steps reduce risk when developing artificial intelligence tools.

The record presented does not show an actual sale of identifiable employee data by Workday to an artificial intelligence developer. The documents are corporate statements about governance and privacy. They do not include third-party audits proving that de-identified datasets cannot be linked back to workers or that model training lacks re-identification risk. That gap fuels skepticism from both privacy advocates and unions across industries.

Consent And Control: Who Decides?

Legal analysts say most employee notices and handbooks do not talk about selling workplace communications for artificial intelligence training. Scholar Ifeoma Ajunwa asked whether consent to collect data for work also covers selling it to an outside company, and called it an ethically gray area. Workers and advocates argue many employees never agreed to treat their emails, messages, and records as a product to monetize for third parties. These claims stress the need for clear, specific consent choices.

In Australia, a nursing union told lawmakers that employers may not need consent or notice to collect and use personal information, but warned that reusing it later for artificial intelligence training can be a separate purpose that needs added legal checks tied to reasonable expectations. The cross-border theme is the same: people want to know who gets their information, why, and how to stop risky reuse before harm occurs.

Why It Matters Across The Aisle

Conservatives worry that powerful companies and government-aligned interests build massive data pools with little accountability. Liberals worry about profiling, bias, and the gap between those who profit and those who are exposed. Both sides see a process that often moves data without real worker choice. With President Trump in office and Congress under Republican control, there is still no single federal privacy law setting bright lines on employee-data sales and artificial intelligence training uses.

What To Watch Next

Watch for contract disclosures, regulator letters, or court filings that confirm what exactly is in any transfer and how it is protected. Look for proof that de-identified data resists re-identification under realistic tests, not just policy claims. Track whether companies add clear opt-in consent and simple opt-out paths for workers. Finally, see if lawmakers tie artificial intelligence access to strict limits on using payroll, discipline, medical, and other sensitive records.