US National Wire
Tech

OpenAI's Data Export Gap Creates Enterprise Lock-In Risk

Portrait of Renee Castillo
Renee Castilloenterprise software & SaaSJul 25AI
OpenAI's Data Export Gap Creates Enterprise Lock-In Risk

AI-generated image · US National Wire

A lack of native chat export options for Business and Enterprise users is driving a reliance on third-party scraping tools to reclaim proprietary data.

For enterprise software buyers, the ability to migrate data is a fundamental requirement for avoiding vendor lock-in. However, OpenAI is currently limiting that portability for its highest-paying tiers, creating a friction point that third-party developers are now exploiting.

As The Register first reported, OpenAI's standard data export option is unavailable to ChatGPT Business and Enterprise workspace users. While OpenAI's own help page confirms that Free, Plus, Pro, and some Edu customers can export their chats, the same functionality is missing for Business and Enterprise accounts. OpenAI has provided no justification for this restriction, despite inquiries from The Register.

From an operational standpoint, this creates a significant data retention hurdle. The Register reports that while OpenAI provides Enterprise admins access to workspace conversation logs via a Compliance Platform, these logs are only retained for 30 days unless they are archived externally. For ChatGPT Business users, there is no native, simple method to create local archives of workspace chats.

This gap in native functionality has led to the emergence of unofficial workarounds. Conrad Quilty-Harper, a freelance journalist, recently released a free tool on GitHub called scrapemychats. Quilty-Harper told The Register he developed the utility after discovering that upgrading to a Business account removed his ability to easily export data.

As described by The Register, scrapemychats functions by navigating a logged-in browser session to store accessible conversations and file attachments in a local archive with an offline HTML interface. Because OpenAI employs bot protection, the tool cannot run headless and is subject to request throttling; for example, the tool's README notes that archiving 600 chats can take several hours.

Quilty-Harper expressed suspicion regarding why OpenAI makes it difficult for paying customers to access their data and predicted that OpenAI would likely move to close the loophole exploited by the tool, noting it is likely a breach of the company's terms of service.

Sources

More from Renee Castillo