To develop effective AI models, substantial amounts of data are essential, and obtaining this data often comes at a significant cost. Despite having access to a wealth of user data from X and extensive information generated by SpaceX, Elon Musk's Grok appears to be in search of alternative data sources. According to insiders who spoke with Bloomberg, the solution might lie in acquiring customer data from defunct companies at a low price.
Facing difficulties in securing cost-effective data sources, Musk’s SpaceX is reportedly exploring the purchase of operational and customer data from startups that have fallen on hard times or gone bankrupt. This need for affordable data has intensified, especially following SpaceX's recent earnings report, where it revealed a staggering $15.8 billion investment in its AI initiatives. In this context, the company’s AI division, xAI, recorded a significant net operating loss of $1.3 billion for the quarter. Musk's ambitions for the Grok chatbot aim to elevate it to compete with established models from other AI entities such as OpenAI and Anthropic. Grok 5, which is set for release before the year's end, is being touted by Musk as potentially reaching artificial general intelligence (AGI) levels, a goal that necessitates considerable training data.
While discussions about these potential acquisitions are reportedly informal and may not culminate in actual agreements, such moves could have serious implications for user privacy. Customers who previously shared their data with now-struggling firms might find that their information ends up in the hands of Musk’s companies.
This quest for data isn’t exclusive to SpaceX; several major tech companies are similarly keen on acquiring data, despite ethical risks and privacy concerns. As AI systems become more integrated into various fields, there has been a marked increase in the demand for industry-specific knowledge, giving rise to a niche market for reselling employee data from failed businesses to AI companies.
For instance, in a recent bankruptcy auction, Google placed a $10 million bid to acquire data from Spirit Airlines, which has ceased operations. While the data doesn’t include passenger details, it encompasses a wealth of information, including 100 million emails, 500 million Microsoft Teams chats related to former employees, operational data, productivity metrics, and legal documents related to employment. Although Google has promised to remove personally identifiable information from the dataset, concerns persist among former employees represented by the Association of Flight Attendants. They lodged an objection in U.S. Bankruptcy Court, arguing that the deidentification processes currently applied may not adequately protect employees' private information.
The union highlighted that even a pseudonymized dataset could potentially reveal sensitive insights about employee grievances, training performance, investigations, and compensation changes. As a result, the court has delayed the decision on approving Google's purchase of the data.
