The AI data firm Micro1, which merchandises corporate data to refine AI model outputs, has pledged a minimum of $1 million to fund AI-skills training for ex-employees of the bankrupt Spirit Airlines. This move is part of a strategy to secure the defunct carrier's corporate data, with the promise aimed at convincing the bankruptcy court that Micro1 can safeguard the information more securely than a rival bid from Google.

In mid-August, Google emerged victorious at an auction, winning the rights to the bankrupt airline's internal operations data and software code. Micro1's CEO, Ali Ansari, argued that factual operational records from a functioning enterprise are crucial for boosting model performance, stating that anonymized real-world business data can recreate authentic operational settings and greatly enhance AI models.

This $1 million proposal was submitted ahead of a September 9 bankruptcy court hearing, which is set to either approve the sale of this data to Google or potentially restructure the disposal process. The plan would afford former Spirit Airlines employees free entry to Micro1's courses, enabling them to launch careers as data contractors. Additionally, the startup has committed to prioritizing the hiring of these laid-off employees for outsourced tasks such as data labeling and model evaluation.

Micro1 had earlier lodged a $12.5 million bid for this data and officially filed these accompanying commitments on Wednesday evening. Google won the bankruptcy auction last month with a $10 million offer. Another enterprise, Mercor, which operates a similar outsourced labor model for AI labs, put forward a $7.5 million bid.

A labor union representing some former Spirit Airlines staff has raised objections to the auction result, and since the court has not yet formally approved the transaction, Micro1 still sees a window of opportunity. Although such scenarios are rare, the court holds the power to overturn the initial decision if it determines this new, later-filed offer is more aligned with the objecting parties' interests.

Beyond the $1 million training expenditure, Micro1 has also promised to de-identify the data, stripping out customer information and sensitive employee records like disciplinary files and collective bargaining materials. The company asserts its approach goes further than Google's acquisition plan, which would only excise customer data but retain employee records. Micro1 further vows to independently scrub personal identification information before transferring data to external parties.

Google declined to comment but has previously stated that any data it receives would undergo rigorous personal identifier removal by a third party, a party slated to be selected by the court within days of the debtor's filing. In recent months, historical corporate Slack logs and ticket data have become highly sought-after assets. This surge is fueled by the expanding use of AI agents capable of handling tasks like tech support responses, clinical trial management, and invoice verification for users. Insiders reveal that Micro1 is currently also eyeing the acquisition of data assets from other bankrupt enterprises.