Google’s acquisition of a huge archive of data belonging to bankrupt Spirit Airlines has raised privacy issues and former employees and their union are protesting the deal. Google won a bankruptcy auction with a $10 million bid for a collection of Spirit’s internal business data that is said to include more than 100 million emails and 500 million Microsoft Teams messages.
The tech giant aims to use the data to help build products and AI models, but former flight attendants are worried employee information that is in the pipeline for the company still could get out even though the company says it is very careful to secure it.
The unusual transaction reflects how valuable corporate data has become in the artificial intelligence race. While technology companies have always relied on publicly available information and licensed datasets to build AI systems, internal business records can provide something else entirely: real-life examples of how employees communicate, make decisions, solve operational problems and manage complex workflows. Spirit’s archive supposedly contains emails and workplace chats as well as documents, spreadsheets, calendars, operational information, software code and other business records.
Google’s bid came in a bidding war and the company initially offered $5 million and AI data company Mercor followed. Google later raised its bid to $10 million and Mercor became the backup bidder to $7.5 million. The sale is a part of bankruptcy recovery and Spirit Airlines’ bankruptcy status. The deal is still subject to bankruptcy court approval and has been delayed by objections from the former employees’ union.
The scale of that archive is what makes this transaction so interesting. The archive contains millions of emails and 500 million Teams messages, and more than 30 million lines of software code according to reports. It also contains information about aircraft operations, revenue management, marketing, human resources, project management, employee productivity, audits and other corporate activity. Such material could give AI developers tangible examples of complex workplace processes that are not easily reproduced using everyday public web data.
But the former employees’ concerns are privacy rather than the commercial value of the data. The Association of Flight Attendants-CWA has protested against the sale, saying that employee-related information might be sensitive even if no information is personal. The union has pointed specifically to disciplinary information, payroll records, internal communications, and other employment-related material.
Google has said that the data it receives won't contain personal information. According to court documents and company statements, a third party will clean the material of personally identifiable information before Google receives it. The company will also be prohibited from attempting to identify individuals from the de-identified information. Passenger profiles, loyalty programs and credit card information are not included in the transaction, the company said.
That assurance, however, has not allayed the concerns. Privacy experts and the union say removing names and other common identifiers may not always eradicate the possibility of knowing people from context. For example, combining the work of people with their job position, location, dates, operational details and conversations may help identify them even though they have their name removed. The union has claimed that certain types of employee information are still sensitive whether they are removed as identifiers.
The dispute also poses a larger question of who controls workplace data once a business shuts down. Employees may interpret emails and internal messages as private workplace talk, but all of those data are kept on the corporate servers. When the company is going through bankruptcy, digital assets can be placed in the estate and sold alongside other business property. Spirit’s case highlights how records from years ago can come with unexpected financial value when a company closes down.
The controversy comes at a time when AI companies are looking for quality data that can be used to enhance models and AI-based software. Real-world enterprise information is particularly appealing because of the examples of practical decision-making and complicated workflows. Spirit’s operational records might help to create AI systems that understand business processes, automate tasks or assist workers.
At the same time, the case could be an important test for privacy rights in companies’ data transactions. If a bankruptcy estate can sell huge archives of employee communications for AI development, workers need to know what protections to expect when their former employer’s digital assets come off the table. The outcome could influence future disputes involving bankrupt companies with valuable databases and workplace communications and proprietary software.
The Spirit Airlines data sale is still under investigation. Google’s $10 million bid shows the growing commercial value of enterprise information, and former employees’ objections are a testament to the challenges of responsibly managing such data.
The fundamental issue is not how much AI can learn from millions of workplace communications but whether the people who created those records can be protected so that the data can be recycled years later.
The dispute is therefore a watershed moment in the way AI development, corporate data and employee privacy are interacting. The bankruptcy court will decide the objections and the safeguards and it will be an important indicator of the extent to which companies can monetise digital records created in the course of everyday work life.
