SpaceX, a company led by Elon Musk, is reportedly exploring novel options to secure cheaper training data for its Grok AI chatbot, including acquiring customer data from financially distressed or bankrupt companies, according to Bloomberg. This strategic move comes as
the company has committed a substantial $15.8 billion to its artificial intelligence endeavors, despite its xAI division reporting a net operating loss of $1.3 billion for the previous quarter. Musk harbors ambitious plans for Grok, aiming for it to catch up
with leading competitors such as OpenAI and Anthropic. The company intends to unveil the Grok 5 model before the end of the year, which Musk claims will achieve Artificial General Intelligence (AGI) – an AI system capable of performing any
intellectual task a human can. This ambitious goal necessitates vast quantities of training data, driving the company to explore unconventional acquisition sources. These discussions, however, remain informal and may not materialize into final agreements. Should such deals proceed, they could raise
significant and widespread concerns regarding data privacy. A company that once provided services to users, but is now facing financial distress, might find itself selling those users' personal data to entities like Musk's firm. This strategy is not exclusive to SpaceX.
As AI agents become increasingly prevalent in workplaces, artificial intelligence laboratories are placing greater importance on specialized knowledge pertinent to each industry sector. This trend has fostered a new niche market centered around financially struggling or failed companies selling their
employee data to AI labs for model training purposes. In a related development, a bankruptcy auction for Spirit Airlines last month saw Google submit a $10 million bid to acquire the defunct airline's data, intended for training its Gemini models. While
the data did not directly include passenger information, the sold package was remarkably comprehensive. It encompassed 100 million company emails and 500 million Microsoft Teams conversations from former employees. Furthermore, it contained aircraft operations data, employee productivity metrics, one million
time card records, 17 million items from Microsoft OneDrive, and hundreds of thousands of employee records, including tax forms, employment contracts, and litigation files. Despite Google's assurance that the data would be de-identified to remove personally identifiable information, this statement was
insufficient to alleviate the concerns of the Association of Flight Attendants. Representing former Spirit employees, the union filed an official objection with the bankruptcy court in New York, arguing that the applied "de-identification" standards only cover consumers, leaving employees' private
data, including travel records and tax documents, vulnerable to exposure.
