OpenAI won't let some customers export their chats, but this tool will
The Register
OpenAI won't let Business and Enterprise ChatGPT users export their own chats. A dev built a free workaround before the loophole gets patched.
Based on reporting by The Register — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Conrad Quilty-Harper had a problem most of us don't think about until it's too late: he upgraded his personal ChatGPT account to a Business one, and only afterward discovered there was no easy way to get his own conversation history back out. OpenAI's export tool works fine for Free, Plus, Pro, and some Edu accounts. Business and Enterprise workspaces? Locked out, with zero explanation on the company's help page and, per The Register's request for comment, zero explanation when asked directly either.
So Quilty-Harper, a freelance journalist rather than a career developer, built scrapemychats and put it up on GitHub this week. The tool logs into your browser session, walks through your ChatGPT conversations one by one, and saves them along with any file attachments into a local archive you can browse offline through an email-style HTML interface. It has to run in a real, logged-in browser window because OpenAI's bot detection blocks headless scraping — and because the company throttles rapid requests, pulling down something like 600 chats can take a few hours, though the process can be paused and resumed.
The existence of this workaround says more about OpenAI's priorities than any statement would. Enterprise admins do get a Compliance Platform for pulling workspace logs, but it only holds onto that data for 30 days unless someone archives it separately, and regular Business users don't get even that. For a company whose product is built on ingesting other people's written and created material at scale, making it hard for paying customers to keep their own written and created material has a certain irony that Quilty-Harper didn't shy away from pointing out.
He's also blunt about the tool's shelf life. He told The Register he expects OpenAI to close the loophole shortly after this story runs, and he suspects using scrapemychats probably violates ChatGPT's terms of service. That hasn't stopped him from publishing it anyway, framing it as a way for people to reclaim content they generated before walking away from the platform for good.
What happens next is the interesting part. If OpenAI patches this quietly, that alone tells you the friction wasn't accidental. If the company adds proper Business-tier export instead, that's a tacit admission that customers were right to be annoyed. Either way, Quilty-Harper says he'd genuinely like someone at OpenAI to explain the original decision — he's still waiting.
My take — AI-written commentary, not fact-checked reporting
I've seen this move before with cloud storage and social platforms: make premium tiers stickier by quietly stripping the exit door, and hope nobody with GitHub access notices. Charging businesses more while giving them less control over their own data isn't an oversight, it's a retention strategy dressed up as a technical limitation. And there's something almost poetic about a company built on scraping everyone else's content getting scraped right back by one annoyed customer with a laptop.
Read more about this at: The Register