Microsoft Blog Post Promoted Training AI on Pirated Harry Potter Books
Microsoft published a developer blog tutorial in late 2024 that instructed users to download pirated copies of the Harry Potter book series to train AI models on Azure. The post, written by a senior product manager, has since been removed, but was cached and shared widely before deletion.
What happened
Microsoft published a developer blog post in late 2024 written by Pooja Kamath, a Microsoft Senior Product Manager, that directed users to obtain pirated copies of the Harry Potter novels through a Kaggle dataset to train Azure-based AI models. The blog post, which was a technical tutorial on adding generative AI to applications via Azure, suggested using the series to create a Q&A system or auto-generate Harry Potter fan fiction. The post was accompanied by an LLM-generated image of two children on a train with a Microsoft logo between them.
Reported impact
The Kaggle dataset, mistakenly marked as 'public domain,' was downloaded approximately 10,000 times before the blog post and dataset were removed from the site. Microsoft did not initially offer a public explanation for the removal, though the incident reignited discussion about the tech industry's relationship with copyright law in generative AI development.
Governance controls
The blog post and associated Kaggle dataset were subsequently removed from Microsoft's website, though both remained accessible via the Internet Archive.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.