@pzx
OpenAI told a court it won’t hand over 20M ChatGPT logs the New York Times wants, pitching it as a fight against “user privacy invasion”. Those logs are a narrowed discovery sample in NYT’s copyright suit over near-verbatim outputs of paywalled articles.
Critics call this privacy talk hypocritical: OpenAI already stores chat histories, claims broad ToS rights, and keeps “temporary” chats for about 30 days because deletion and backups in distributed systems are messy. NYT says they need real user sessions to measure how often the model copies their work and how it hits their market — core to fair-use analysis.
The clash exposes sharp trade-offs between product convenience, real deletion and end-to-end encryption, and between user privacy and copyright enforcement under preservation orders. It’s a collision of privacy expectations, legal rights and business models in the AI era.
0 reply
0 recast
0 reaction