{"version":1,"type":"story","url":"https://digestai.news/story/ai-labs-buy-data-from-dead-startups-for-training-models","json":"https://digestai.news/story/ai-labs-buy-data-from-dead-startups-for-training-models.json","markdown":"https://digestai.news/story/ai-labs-buy-data-from-dead-startups-for-training-models.md","slug":"ai-labs-buy-data-from-dead-startups-for-training-models","headline":"AI labs buy data from dead startups for training models","summary":"AI labs are purchasing data from defunct startups to train models, as public and synthetic datasets prove insufficient. Companies like SimpleClosure and Sunset broker these deals, selling codebases, Slack archives, emails, and workspace data—often for tens of thousands of dollars. SimpleClosure’s Asset Hub platform has closed over 150 deals, expecting to return $10 million to startups by year-end, with 47% now involving code or data sales. Buyers include foundation AI labs, vendors creating simulated workspaces, and established tech firms like Google, which paid $10 million for Spirit Airlines’ 100 million emails and 500 million Teams messages during its bankruptcy in August. Demand surged last year as AI training hit dataset limits, with labs like Surge AI using real-world archives to build reinforcement-learning environments for AI agents to practice tasks like debugging or drafting legal briefs. Pricing remains speculative, based on estimates of useful tasks per data volume, but deals are growing routine despite legal and ethical concerns. Critics warn of privacy risks and unclear profit-sharing, as proceeds often go to investors rather than employees. Legal experts describe the practice as a ‘Wild West’ with insufficient licensing or consent frameworks.","keyPoints":["SimpleClosure’s Asset Hub sold over 150 startup data deals, expecting $10M in returns by 2026 end","Google paid $10M for Spirit Airlines’ 100M emails and 500M Teams messages in bankruptcy auction","AI labs buy Slack, email, and code archives to train models, but pricing and legal risks remain unclear"],"whyItMatters":"This trend highlights how AI labs exploit discarded startup data to fill training gaps, raising ethical and legal questions about consent and ownership. It also signals a growing ‘data undertaker’ industry, though uncertainty over value and privacy risks persists.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":["SimpleClosure","Sunset","Surge AI","Google","Spirit Airlines","Mercor"],"models":[],"people":["Keelan Miskell","Dori Yona","Brendan Mahony","Nick Heiner","Greg Ewing","Ifeoma Ajunwa"]},"firstPublishedAt":"2026-10-13T09:00:00Z","updatedAt":"2026-10-13T09:00:00Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"MIT Technology Review AI","title":"AI labs are buying data from dead startups to train their models","url":"https://technologyreview.com/2026/10/13/1145755/ai-labs-buying-data-from-dead-startups-to-train-models","publishedAt":"2026-10-13T09:00:00Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"AI labs buy data from dead startups for training models\", 13 October 2026, https://digestai.news/story/ai-labs-buy-data-from-dead-startups-for-training-models","publisher":"Digest AI","title":"AI labs buy data from dead startups for training models","datePublished":"2026-10-13T09:00:00Z","url":"https://digestai.news/story/ai-labs-buy-data-from-dead-startups-for-training-models"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}