MORE THAN 120 U.S. LAWMAKERS CHALLENGE GOOGLE OVER PROPOSED PURCHASE OF 600 MILLION SPIRIT AIRLINES EMPLOYEE COMMUNICATIONS

More than 120 members of the United States Congress are challenging Google’s proposed $10 million purchase of an enormous archive of internal data from collapsed Spirit Airlines, warning that the transaction could transfer approximately 100 million emails, 500 million Microsoft Teams messages and years of sensitive employee information into the hands of one of the world’s largest artificial-intelligence companies without adequate protection for thousands of former workers whose communications were created as a condition of their employment.

The intervention came in a newly dated congressional letter released Thursday and led by Democratic Senator Elizabeth Warren of Massachusetts and Representative Steven Horsford of Nevada. A total of 121 senators and representatives signed the letter to Google Chief Executive Sundar Pichai and Spirit Chief Executive Dave Davis, urging the companies to exclude employee-related information from the transaction as extensively as possible and demanding stronger, enforceable safeguards for material that cannot be removed. The lawmakers said the proposed transaction enters largely uncharted territory because Google has acknowledged that the Spirit business dataset could be used to improve its products and train artificial-intelligence models, turning the internal digital history of a defunct airline into a potentially valuable AI-development asset.

Google won an auction for the Spirit data in August with a $10 million bid after the airline ceased operations and its remaining assets began being sold through bankruptcy proceedings. Court filings and subsequent reporting show that the package contains far more than ordinary corporate documents. In addition to approximately 100 million emails and 500 million Teams messages, it includes spreadsheets, calendars, operational records, marketing material, employee productivity information, software code and other internal business data accumulated during Spirit’s years of operation. Congressional leaders said the material also includes employee records, timecards, payroll and tax information and employment contracts, raising concerns that highly personal workplace information could be transferred even if conventional identifiers such as names and Social Security numbers are removed.

The central dispute is therefore not whether Google bought a conventional customer database, because passenger profiles and frequent-flyer information are excluded from the proposed transaction. It is whether thousands of former Spirit employees should have their workplace communications and employment records incorporated into a commercial dataset that Google intends to use for product development and artificial intelligence. Google and Spirit have said personally identifiable information will be removed before the transfer, and the transaction includes requirements that Google not attempt to re-identify individuals. Lawmakers and employee unions argue that de-identification does not necessarily resolve the problem because the contents of workplace communications can remain sensitive even when obvious personal identifiers are stripped away.

Warren and Horsford’s letter argues that Spirit employees did not create those records with the expectation that they would later be sold to train another company’s AI systems. Workers used corporate email, Microsoft Teams, payroll systems and other databases because their jobs required them to do so, meaning they had little practical choice about whether the information was created or retained. The lawmakers contend that bankruptcy should not transform such information into an unrestricted commercial asset without meaningful consideration of worker privacy, particularly when the archive may include medical accommodation requests, disciplinary records, workplace complaints, tax information, compensation details and communications involving labor relations.

The sale has already attracted significant opposition from labor organizations representing Spirit employees and other airline workers. The Association of Flight Attendants-CWA has objected to the transaction, arguing that the protections negotiated around customer information are stronger than those covering employee data. The Air Line Pilots Association and the Allied Pilots Association, which represents American Airlines pilots, have also raised concerns, particularly about training and safety information that could be contained in the archive. The pilots’ unions argue that aviation safety systems depend partly on employees being willing to report incidents, mistakes and operational concerns candidly, and that workers may be less willing to participate in such programs if they believe confidential records could eventually be sold to outside corporations.

Those objections give the dispute implications beyond ordinary privacy concerns. Airlines operate under extensive safety-reporting systems in which pilots, cabin crew, mechanics and other employees are encouraged to document potential risks and operational problems. Some of those programs depend on confidentiality protections intended to encourage workers to report mistakes or near misses without fear that the information will later be used against them. If such records are included in the Spirit dataset, labor organizations say the sale could undermine trust in voluntary safety programs across the aviation industry, even if Google has no intention of identifying individual workers.

Google has defended the proposed acquisition by emphasizing that it will not receive personally identifiable information. The company has said that an independent third party will scrub the dataset before transfer and that the information will be de-identified. Google also publicly confirmed when the auction became known that it acquired part of Spirit’s enterprise dataset because the information could help improve its products and AI models. That confirmation is significant because it removes much of the uncertainty surrounding the commercial purpose of the acquisition: the internal records are valuable not simply as archived airline documents but as real-world enterprise data that can be used in developing and testing artificial-intelligence systems.

The Spirit archive is potentially attractive for that purpose because it represents decades of real corporate activity rather than artificial or publicly available training material. Internal emails and Teams conversations capture how employees communicate, solve problems, make decisions, coordinate schedules and respond to operational disruptions. Spreadsheets, presentations, financial databases, customer-service processes, software code and business records can similarly provide examples of how a large organization actually functions. For companies developing enterprise AI tools capable of summarizing communications, automating workflows, assisting employees and analyzing corporate data, such a dataset could have considerable value.

The transaction therefore highlights a broader transformation taking place as artificial intelligence increases the economic value of information that companies previously regarded mainly as operational records. Corporate archives that once represented storage costs or legal liabilities can now become valuable training material for technology companies seeking vast quantities of realistic business data. The collapse of Spirit Airlines has created an unusual test case because the company is no longer operating and its bankruptcy estate is legally required to seek value from remaining assets for creditors, while former workers have little direct influence over how those assets are sold.

Spirit Airlines shut down in May after failing to recover from its second Chapter 11 bankruptcy in two years. The ultra-low-cost carrier had struggled for years with mounting debt, higher operating costs, aircraft problems and intense competition, while a proposed $3.8 billion merger with JetBlue Airways was blocked by a federal judge in 2024 on antitrust grounds. The failed merger eliminated what had been seen as a potential route out of Spirit’s financial difficulties, and the airline later returned to bankruptcy before eventually ceasing operations. Since then, its aircraft, airport assets, intellectual property and other remaining property have been sold or evaluated through the bankruptcy process.

Google emerged as the winning bidder for the internal data archive during an August auction, offering $10 million. AI recruiting company Mercor submitted a $7.5 million bid and was designated as a backup bidder, while reports later indicated that another AI company, Micro1, attempted to offer a higher amount after the original bidding deadline. The fact that multiple AI-related businesses were interested in the dataset reinforces the commercial value being attached to real-world corporate communications as machine-learning material.

The proposed transfer still requires final approval from the U.S. Bankruptcy Court for the Southern District of New York. A court-appointed consumer privacy ombudsman has recommended approval after Google and Spirit modified the transaction to strengthen consumer protections and remove passenger databases. The companies also engaged third-party data company Tonic.ai to de-identify information before it reaches Google. The ombudsman concluded that those changes substantially reduced privacy risks for consumers, but the review focused primarily on passenger information and did not resolve the separate objections concerning employees.

That distinction has become a central argument for unions and lawmakers. Spirit’s passenger data is being excluded or tightly protected, while former employees fear their work communications, employment histories and internal records remain part of the package. Critics say this creates a situation in which customers who voluntarily bought airline tickets could receive stronger privacy treatment than employees who were required to use Spirit’s digital systems to perform their jobs.

The congressional letter attempts to close that gap by asking Google and Spirit to remove employee information wherever technically possible and establish enforceable protections where removal cannot be accomplished. The lawmakers have not demanded that the entire transaction be prohibited, nor have they argued that Google should be barred from buying every form of Spirit business data. Their position is that employee communications and records deserve a separate privacy analysis rather than being treated simply as another category of bankrupt corporate property.

That approach reflects the complexity of the dataset itself. Some Spirit records may have little connection to individual workers and could be commercially useful without creating significant privacy risks. Aircraft maintenance databases, generic operational processes, software code or marketing systems may be relatively straightforward to transfer after appropriate review. Internal emails and Teams conversations are different because they can contain references to identifiable individuals, personal circumstances, medical conditions, workplace conflicts, disciplinary issues, union activity and confidential safety concerns even after names and account identifiers are removed.

De-identification is not a perfect technical solution in such circumstances. A message may not contain a worker’s name but could identify the person through a job title, flight number, location, date, supervisor or description of a specific incident. Large language models and modern analytical systems are particularly capable of connecting contextual information across enormous datasets, meaning traditional anonymization techniques may not always provide the degree of protection workers expect.

The lawmakers’ concerns therefore go beyond the possibility of Google deliberately attempting to discover an employee’s identity. They are also questioning whether sensitive information should be transferred at all if it could be incorporated into systems that learn patterns from workplace communications. Once data has been transferred and processed through machine-learning pipelines, reversing that use can become significantly more complicated than simply deleting a conventional database.

The proposed acquisition also illustrates the weakness of American employee-data protections compared with some other areas of privacy law. The United States still lacks a single comprehensive federal privacy law governing all employee information. Instead, protections come from a mixture of state statutes, employment law, collective-bargaining agreements, sector-specific rules and contractual obligations. California and several other states have expanded privacy rights, but protections vary significantly depending on where an employee lived or worked and what kind of information is involved.

Bankruptcy adds another layer of complexity because companies in financial distress are expected to maximize the value of their assets for creditors. Digital information can therefore be sold even when the people described by that information never viewed themselves as participants in a commercial transaction. Courts have dealt with similar questions involving customer lists, health records and loyalty programs, but the emergence of generative AI gives workplace archives a new type of commercial value that bankruptcy law was never originally designed to address.

The Spirit case could therefore influence how future bankruptcies treat digital employee records. Modern corporations produce enormous quantities of communication through email, Slack, Microsoft Teams, Zoom transcripts, cloud documents, internal ticketing systems and collaboration platforms. If those archives are treated as transferable assets whenever a company fails, employees may discover that workplace conversations can survive long after both their jobs and their employers have disappeared.

That possibility has attracted particular concern from organized labor. Flight attendants and pilots argue that employees should not be forced to choose between participating honestly in internal systems and worrying that years later their communications could become training material for an unrelated technology company. The concern is especially acute in aviation because confidential reporting and internal discussion can directly affect operational safety.

Google’s role also brings additional scrutiny because of its position in artificial intelligence. Alphabet has invested heavily in Gemini and other generative-AI technologies while integrating machine-learning tools throughout Google Search, Workspace, Cloud and other products. Enterprise AI has become an increasingly important area of competition as technology companies develop systems capable of working with email, spreadsheets, corporate documents and internal communications. A real-world archive from a large airline therefore offers a different kind of training environment from material scraped from the public internet.

There is nothing inherently unlawful about using properly acquired business data for AI training, and the congressional letter does not establish that Google has violated privacy law. The dispute concerns whether existing legal protections are adequate when data created by workers for one employer is sold to another company for a fundamentally different technological purpose. Google’s promise to remove personally identifiable information addresses one part of that problem, but lawmakers argue that it does not resolve questions of consent, confidentiality and ownership.

The $10 million purchase price is comparatively small in the context of Google’s finances but significant within Spirit’s bankruptcy estate. It demonstrates that a dataset containing internal communications and corporate operations can itself become a valuable asset even after the company that created it has stopped doing business. The competing bids from other AI companies suggest that Spirit’s archives are being valued partly because they offer something increasingly difficult to obtain: a large, coherent collection of genuine workplace data covering the internal operations of a major corporation over many years.

For creditors, accepting the highest qualified bid is consistent with the objective of recovering as much value as possible from Spirit’s remaining assets. For employees, however, that logic can appear fundamentally different because the value being monetized was created partly through their everyday communications and employment records. They may receive none of the proceeds while bearing the privacy consequences of the transfer.

That tension is likely to become more common as artificial intelligence changes the economics of corporate information. Companies previously retained records primarily for operational, regulatory or legal reasons. Those same records may now be commercially attractive because machine-learning developers require large, high-quality datasets that reflect how real organizations function. Bankruptcy can expose that value in particularly stark form because nearly everything owned by a failed company may be considered for sale.

Congressional involvement could pressure Google and Spirit to negotiate stronger safeguards before the bankruptcy court makes a final decision. Possible measures could include excluding entire categories of employee records, removing confidential safety-reporting information, imposing tighter restrictions on AI training uses, creating independent auditing requirements or allowing workers and unions greater visibility into what is being transferred. The lawmakers’ letter does not prescribe every technical solution, but it makes clear that simply removing names from hundreds of millions of communications is not considered sufficient by the signatories.

The scale of the congressional coalition also increases the political significance of the case. The letter is backed by 121 lawmakers, meaning concern has moved well beyond a single senator, representative or labor union. Warren has long been a prominent critic of the concentration of power among large technology companies, while Horsford represents Nevada, where Spirit had a significant employment presence through its operations in Las Vegas. Their joint involvement places both national privacy policy and the direct interests of former workers at the center of the dispute.

The case is unfolding as Congress continues debating broader rules governing artificial intelligence and data ownership. Lawmakers have questioned how technology companies obtain training material, whether users should have greater control over information used by AI systems and whether existing privacy laws are capable of addressing machine-learning models that can process billions of records. The Spirit transaction offers an unusually concrete example because the source and scale of the data are publicly known and the people whose information is involved can be identified as a specific workforce.

Google and Spirit will now have to persuade the bankruptcy court that the protections built into the transaction are sufficient. Google has emphasized that it will not receive personally identifiable data and that a third party will scrub the records before transfer. Supporters of the deal can also point to the exclusion of customer databases and the independent privacy review as evidence that the transaction has been modified in response to concerns.

Workers and lawmakers are arguing that the remaining problem is not merely identification but confidentiality. A former employee may reasonably consider an internal discussion about medical leave, workplace harassment, disciplinary action or a confidential safety issue private even if a name is removed from the record. The information was created inside a closed employment environment and may reveal sensitive circumstances through context alone.

The bankruptcy judge will ultimately have to weigh those concerns against Spirit’s obligation to monetize its remaining assets, while considering the safeguards attached to the proposed sale. The outcome could determine whether the Google transaction proceeds largely as planned, is modified further or faces another delay while additional employee protections are developed.

For former Spirit employees, the dispute has emerged months after they already endured the collapse of their employer and the loss of their jobs. They are now confronting the possibility that the internal digital record of their working lives could outlast the airline itself and become part of the development of another company’s artificial-intelligence products. That is precisely the scenario Warren, Horsford and the other lawmakers say requires stronger safeguards before the data leaves Spirit’s control.

The controversy is therefore larger than the $10 million price attached to the proposed sale. It raises a fundamental question about whether workplace data should be treated like any other corporate asset when a company fails, particularly when advances in artificial intelligence suddenly make decades of emails, chats and internal records commercially valuable in ways that neither employees nor employers could have anticipated when much of the information was created.

The immediate issue before the bankruptcy process is whether Google can acquire Spirit’s de-identified business archive while providing sufficient protection for former workers. The broader issue is whether American privacy and bankruptcy law are prepared for an economy in which the digital history of a failed company can become valuable raw material for artificial intelligence. More than 120 lawmakers are now signaling that, in their view, the protections offered so far do not adequately answer that question, and the court’s treatment of Spirit’s 600 million workplace communications could establish an important precedent for how employee data is handled in future corporate collapses.