The insatiable appetite of artificial intelligence models for vast quantities of high-quality data has ignited a new and rapidly expanding industry: the acquisition and licensing of corporate and individual digital workflows. This burgeoning market, characterized by significant financial transactions and fueled by the need for ever-more sophisticated AI, is quietly transforming how companies value their internal records and raising profound questions about data privacy, consent, and the future of work. In recent months, numerous tech publications and industry observers have noted a surge in unsolicited pitches from data acquisition firms to media companies, freelancers, and even academic institutions, seeking access to everything from old documents and emails to Slack messages and call transcripts. These overtures, often involving substantial sums, signal a shift in how AI development is being financed and resourced.
The Influx of Data Acquisition Pitches
Journalists and content creators have become a primary target for these data-seeking entities. One journalist recounted receiving an email from an individual named Nick, who described a new trend: "AI labs such as Anthropic and OpenAI are paying media companies such as Wired $350K to $1.8M as a one-time fee for a copy of their old work records (documents, emails, slack messages, tickets, etc.) for training better models." Another similar pitch, which broadened the scope to include call transcripts, escalated the potential remuneration to $2 million, with the sender, Cora, noting that "Public data is exhausted, and the labs are spending over $45B on private data to build better models."
These figures, while anecdotal in origin, point to a significant financial investment. The underlying premise is that publicly available datasets, once the bedrock of AI training, are becoming increasingly insufficient and saturated. To overcome this limitation and to develop more nuanced, context-aware, and specialized AI models, leading AI laboratories are turning to proprietary and "messy" real-world data that reflects actual human workflows. This shift has given rise to a new breed of companies dedicated to sourcing, cleaning, and licensing this valuable information.
The Rise of Data Acquisition Startups
The landscape of AI data acquisition is rapidly evolving, with several companies emerging as key players. Firms like Handshake, Micro1, and Mercor are at the forefront of this industry, often described as a "modern gold rush." These companies specialize in brokering deals between data-rich organizations and AI developers.
Handshake, for instance, confirmed to one reporter that it licenses professional documents across various fields to enable AI labs to "test and train how models handle real knowledge work." The company also revealed that a significant portion of its business involves acquiring graduate students’ papers and coursework, highlighting the breadth of its data-sourcing activities. Similarly, a freelance writer shared an experience with Handshake, which reportedly offered her $1,000 for the data associated with a single story, a proposition she ultimately declined.
Another emerging segment of this market involves companies that facilitate the "winding down" of businesses. Startups like Sunset specialize in handling the complete data divestiture process for companies ceasing operations. Their service includes "scrubbing and anonymizing the data so the company people and any identifying details are stripped out, then brokering the deal with the labs." This service addresses the operational and legal complexities associated with data disposal for companies facing closure, positioning them as intermediaries in the data value chain.
Industry Leaders and Their Business Models
Mercor, founded in 2023, has quickly established itself as a dominant force in the data acquisition sector. Its three co-founders reportedly became billionaires at the age of 22 in the preceding year, underscoring the rapid ascent and lucrative nature of this industry. Mercor’s business model, according to CEO Brendan Foody, evolved from an AI recruiting firm. By gaining insights into the AI industry’s pain points, the company identified a critical bottleneck: the scarcity of high-quality, real-world data. Foody emphasized that "Big AI has two bottomless needs: compute and data," drawing a parallel between the publicly visible investments in computing power and the less visible, yet equally crucial, efforts in data gathering. Mercor claims partnerships with nearly all major AI labs, a statement that AI companies themselves have responded to with more measured acknowledgments of the data market’s significance.
Micro1, founded in 2022, is another prominent player. CEO Ali Ansari projects that the current annual expenditure of $40 billion to $50 billion on AI data by labs, enterprises, and robotics companies will skyrocket to over a trillion dollars within the next two to four years. Micro1 leverages acquired data both for developing its own AI models and for resale to other major AI firms. Ansari noted the intense demand, stating that "Pretty much everybody’s trying to get into the data space now because there’s so much demand," citing the significant proportion of Y Combinator startups classified as data companies.
The Privacy Conundrum and Ethical Considerations
Despite the rapid growth and financial allure of the AI data acquisition market, significant ethical and privacy concerns loom large. A pivotal case that brought these issues to the forefront involved the bankruptcy proceedings of Spirit Airlines. The airline sought to sell its digital assets, a trove of 80,000 email accounts, 500 million internal messages, 20.6 million Microsoft SharePoint files, and a million time-card records. Google initially appeared to secure the assets with a $10 million bid, but Micro1 subsequently topped it by $2.5 million, leading to a dispute over the winner.
The sale was placed on hold due to objections from the flight attendants’ union, which alleged that the transaction involved their private information. This objection highlighted a fundamental tension: while data acquisition firms often assert that they rigorously strip Personally Identifiable Information (PII) from datasets, critics question the efficacy and completeness of such anonymization processes.
Mayu Tobin-Miyaji, general counsel for the Electronic Privacy Information Center (EPIC), which filed an amicus brief in support of the flight attendants, expressed skepticism: "It’s not clear that anonymizing can fully get rid of privacy issues. Emails and Teams messages contain unstructured data, which is harder to root out. There may be messages where employees talk about doctor’s appointments or family issues." The unstructured nature of much of this workflow data makes it particularly challenging to guarantee complete de-identification, raising the specter of sensitive personal details being inadvertently exposed.
Regulatory Gaps and the "Wild West" Atmosphere
The rapid expansion of the AI data market has outpaced the establishment of clear regulatory frameworks and industry standards. Gordon Ritter, a venture capitalist actively engaged with these issues, observed, "We haven’t set the norms and the standards for data acquisition, and it’s not being done appropriately. The AI industry is moving far faster than our ability to actually set the standards, and it’s partially intentional." This "Wild West" atmosphere allows for aggressive data acquisition strategies, often without explicit and informed consent from the individuals whose data is being collected.
The core ethical dilemma lies in the inherent purpose of this data acquisition. The very workflows being purchased – documents, emails, messages – are often generated by individuals performing their jobs. The ultimate goal of training AI models on this data is frequently to automate or even replace the human labor that produced it. As the president of the flight attendants’ union articulated, "This is really only data that became valuable in the age of AI in order to train models to replace jobs and commodify other portions of working in a work environment." This creates a paradoxical situation where workers’ contributions are leveraged to potentially undermine their own future employment prospects.
The Future of Corporate Data and Worker Implications
The ongoing trend suggests a future where the ongoing sale of workflow data could become a standard revenue stream for corporations, potentially surpassing advertising revenue. Micro1’s CEO, Ali Ansari, anticipates this shift, stating, "I think it’s going to become more common than ads as a revenue line item for companies." While proponents argue that this will lead to greater efficiency and abundance through advanced AI, a significant segment of the workforce and privacy advocates remain apprehensive. The idea that every professional interaction and every piece of work generated could be funneled into AI models for commercial exploitation raises deep-seated concerns about data ownership and the evolving nature of employment.
Despite these reservations, Ansari remains optimistic, suggesting that workers will eventually understand the benefits of improved AI models that can "make their lives easier." This perspective, however, relies on the assumption that the "upside" of AI-driven abundance will outweigh the immediate concerns about privacy, consent, and job security.
The emergence of companies like Tenkara, which aims to boost the quality of its AI model for running office operations in manufacturing companies, further illustrates the broad application of this data acquisition strategy. While the connection between a publication’s editorial archives and manufacturing operations might seem tangential, it underscores the pervasive search for any data that can refine AI’s ability to understand and replicate complex human tasks.
The current trajectory of AI data acquisition presents a complex interplay of technological advancement, economic opportunity, and significant ethical challenges. As the industry continues its rapid expansion, the need for robust regulatory oversight, transparent data handling practices, and a clear understanding of individual rights will become increasingly critical to navigate this new frontier responsibly. The ongoing debate surrounding the value and ownership of human-generated data in the age of AI is far from over, and its resolution will profoundly shape the future of work and society.
