Google Wants a Dead Airline's Data

AI Summary

What this is: Google won a bankruptcy auction for Spirit Airlines’ internal business data, planning to use it to train AI. A flight attendants’ union has now stalled the deal in court.

Who it’s for: Anyone curious about where AI training data actually comes from, and what happens to a company’s records when it goes bankrupt.

The key idea: The sale excludes passenger data. It does not exclude employee data, and that’s exactly what a union representing Spirit’s flight attendants is fighting in bankruptcy court.

Skip if: You want general background on how AI models get trained. Start with our training data glossary entry instead.

Spirit Airlines stopped flying on 2 May 2026, its second bankruptcy filing in two years, with about $8.1 billion in debt and roughly 17,000 people out of a job. Three months later, what’s left of the company is still generating money, just not from flights.

On 14 August, Google won a bankruptcy auction for Spirit’s internal business data with a $10 million bid, beating an offer of $7.5 million from Mercor, an AI-focused recruiting firm. Google says it wants the records to help train its AI models. A union representing Spirit’s flight attendants filed a court objection days later, and the judge has now pushed the approval hearing back.

Table of Contents

What is Google buying?

The court filing lists roughly 100 million emails, 500 million Microsoft Teams messages, about 17 million OneDrive files, and more than 20 million SharePoint items. It also includes 516 code repositories holding around 30 million lines of Spirit-built software, plus employee records dating back to 1986 and 7.2 billion competitor flight-pricing records the airline had collected over the years.

Asset	Approximate volume
Employee emails	100 million
Microsoft Teams messages	500 million
Lines of custom code	30 million
OneDrive / SharePoint files	37 million+ combined
Competitor flight-pricing records	7.2 billion
Winning bid	$10 million (Google)

What’s excluded from the sale?

Spirit’s 97.5 million passenger profiles and roughly 50.2 million Free Spirit loyalty records are not part of the deal. Google says a third party will strip any remaining personally identifiable information out of the rest before the data reaches Google, a process bankruptcy filings call deidentification.

That still leaves 100 million emails written by Spirit employees and 500 million internal chat messages, none of it customer data.

Why is a union fighting this?

The Association of Flight Attendants-CWA filed a formal objection, arguing that the sale’s privacy language protects passengers but says almost nothing about employees. The union points to disciplinary records, training deficiencies, medical and accommodation requests, payroll history, timekeeping data, and internal Teams messages as material still on the table.

The union’s core argument comes down to scale. Spirit employed a specific, countable number of flight attendants, and the union argues that in a dataset that size, stripping out names doesn’t guarantee anonymity: crew base assignments, shift patterns, and other operational details could still narrow a record down to one identifiable person or a small identifiable group.

What is the union asking the court to do?

The AFA-CWA’s filing asks for one of two outcomes. Either the court rejects the sale unless every record connected to flight attendants, including training files, timecards, payroll data, and any Microsoft 365 content, is carved out entirely, or the court attaches conditions: independent review of any labor or disciplinary material before transfer, and a binding rule against using the data to profile identifiable groups of flight attendants.

What happens next?

The hearing that was set for 19 August has been pushed to 9 September, giving the court time to weigh the union’s objection before ruling on the sale. Nothing has changed hands yet. Mercor remains the backup bidder at $7.5 million if the Google sale falls through.

What AI can’t do here

An algorithm cannot decide whether deidentification actually works for a specific, small workforce. That’s a factual question for the court, informed by people who understand Spirit’s staffing structure.
Deidentifying the data doesn’t answer the union’s real question: whether an employee’s work messages become a company asset to sell off the moment the employer goes bankrupt.

Know Where the Data Comes From

Human curated and edited, for beginners who want a finger on the pulse without the hype.

Free forever. Unsubscribe anytime.

Common questions about the Google-Spirit data sale

Did Google buy passenger data?
No. Spirit’s passenger profiles and Free Spirit loyalty records are explicitly excluded from the sale.

Is the deal final?
No. Google won the bankruptcy auction on 14 August, but a federal judge still has to approve the sale. The approval hearing was rescheduled to 9 September after a union objection.

Who is objecting, and why?
The Association of Flight Attendants-CWA, which represented Spirit’s flight attendants. It argues employee records, including disciplinary, medical, and payroll data, aren’t adequately protected and could still be traced back to individuals even after deidentification.

What would Google do with the data?
Google has said it plans to use the deidentified records to help train its AI models.

What happens if the sale is rejected?
Mercor, an AI recruiting firm, is the backup bidder at $7.5 million and would be next in line.