Character.ai’s bankruptcy has thrust a largely overlooked risk of the AI boom into view: what happens to the intensely personal data users share with chatbots when companies fail. The startup agreed to sell some assets to Disney for $20 million, but user data remains on the auction block, prompting a coalition of 40 U.S. state attorneys general to seek safeguards for children and privacy. Legal experts note that many privacy policies carve out exceptions in bankruptcy, potentially allowing buyers broad latitude to reuse data for training models, targeted ads and other purposes.
The case underscores concerns that an AI bubble—fueled by high compute costs and uncertain profits—could trigger a wave of mergers and liquidations, scattering sensitive conversations across secondary markets. While the EU imposes tighter constraints under GDPR, bankruptcies can still transfer datasets so long as buyers honor original policies, which often permit expansive use. Consumer advocates urge users to delete data where possible and reconsider sharing sensitive information with AI services whose longevity—and stewardship—are far from guaranteed.
Related article:
