Conversations between former employees can earn defunct startups hundreds of thousands of dollars. There is a feeling of “gold rush”.
Shutdown startups are now selling their business data, including emailsmessages on technical forums and tickets support, for amounts that can reach hundreds of thousands of dollars, with the aim of helping to train artificial intelligence models.
It is reported that companies specializing in closing startups are now helping their founders raise some last minute moneymonetizing your internal communications.
While large LLMs like ChatGPT were initially trained with internet-accessible data such as books, news, Wikipedia, and Reddit threads, newer models, in particular AI agentsrequire more complex data, which reflects the form how the work is actually done.
This training often takes place in so-called “reinforcement learning gyms” or RL gyms (reinforcement learning environments), says . It is about simulated environments, with real data from companieswhich allow AI agents to practice performing office tasks.
This type of training data quickly became very profitable. According to , Anthropic directors discussed the possibility of spending up to billion dollars in RL gyms.
As companies that help startups closer, dealing with issues such as salaries, taxes and settlements with investors, are now entering this business. For example, recently introduced a new product called Asset Hubaimed at helping startups monetize their data.
The platform allows companies license elements as source codeas well as workplace data, including documentsworkflows, and internal communications, such as technical support messages and emails.
According to information available on its website, SimpleClosure helps companies determine what data can be sold, to assess your value and to process them in order to remove personally identifiable information.
According to Forbes, in the last year, SimpleClosure processed around 100 of these agreements, with payments that varied between 10,000 and 100,000 dollars per company.
“There’s a feeling of a gold rush by these companies trying to get their hands on real-world data,” SimpleClosure CEO Dori Yona told Forbes.
Still, some privacy advocates are raising concerns about what this means for workers whose data may be included in these datasets.
“I think that the privacy issues here are quite significant“, declared to Forbes Marc Rotenbergfounder of the Center for AI and Digital Policy.
“Worker privacy is a central concern, especially as people have become so dependent on these new internal messaging tools… This is not generic data. They are identifiable people“.
