NEW YORK, August 18, 2026, 20:05 EDT
- A bankruptcy court moved the approval hearing from August 19 to September 9.
- Google offered $10 million for Spirit Airlines’ internal business data.
- A cabin-crew union says preserved links could expose workers after de-identification.
A U.S. bankruptcy court delayed approval of Google’s proposed $10 million purchase of Spirit Airlines’ internal data. The September 9 hearing will follow a union challenge over employee privacy.
The 21-day delay tests a growing market for corporate records used in artificial intelligence. Google, part of Alphabet (NASDAQ:GOOGL), wants the files for product development and model training.
The dispute turns on linkage, not just deletion. Spirit says personal identifiers will be removed. The Association of Flight Attendants-CWA says retained connections across datasets could reveal individuals or small groups.
| Bidder | Offer | Stated use | Status |
|---|---|---|---|
| $10 million | Product development and AI model training | Selected bid; court approval pending | |
| Mercor | $7.5 million | Not disclosed in cited reports | Lower competing bid |
| Google premium | $2.5 million | 33.3% above Mercor’s bid | Calculated from reported bids |
Google’s offer topped Mercor’s by one-third. Yet price is only one test. The court must also weigh safeguards for a workforce whose communications became a saleable bankruptcy asset.
The proposed collection is unusually broad. A bankruptcy filing describes roughly 100 million emails and 500 million Teams messages. It also covers documents, workflows and software code, according to a report on the filing.
| Data category | Reported scale | Proposed treatment | Open issue |
|---|---|---|---|
| About 100 million messages | Personal identifiers removed | Links may reconnect records | |
| Teams messages | About 500 million messages | Personal identifiers removed | Small groups may remain identifiable |
| Spreadsheets and calendars | Not disclosed | Included where business-related | Context may expose roles or events |
| Workflows and code | Not disclosed | Included | Scope and downstream use |
| Customer data | Not applicable | Excluded | No challenge reported |
Those two message stores alone total about 600 million records. That figure is preliminary because the filing’s inventory may change. The $10 million bid also covers other assets, so it does not set a clean per-message price.
Spirit argues the package excludes customer information and personally identifiable employee data. The union’s objection targets residual structure. A name can disappear while timing, team membership and repeated contacts still narrow identity.
| Date | Development | Why it matters |
|---|---|---|
| May 2, 2026 | Spirit ceased operations | Its internal systems became liquidation assets |
| August 14, 2026 | Sale filing reported as submitted | Outlined the proposed dataset and safeguards |
| August 17, 2026 | Google’s $10 million bid reported | Established the proposed buyer and AI use |
| August 19, 2026 | Original approval hearing | Now adjourned |
| September 9, 2026 | Rescheduled hearing | Court can consider the union’s objection |
The airline ceased operations on May 2 after financial strain ended its turnaround effort. Its flight-attendant union then urged officials and the bankruptcy court to prioritize workers.
That history changes the commercial balance. Spirit needs to maximize recovery for creditors. Former workers, however, cannot revise years of routine messages created under different expectations.
The hearing delay does not reject the sale. It gives the court time to examine the objection and any revised safeguards. Google and Spirit had not announced new restrictions by the Reuters publication time.
Risks: Approval could still expose patterns that identify workers despite redaction. Rejection may reduce recovery for the bankruptcy estate. Any tighter order could also limit the dataset’s value for AI training.
The September hearing may set no binding national rule. Still, its terms could become a practical template. More failed companies now hold workplace archives that AI developers can value after the original business disappears.
