2 Comments
User's avatar
тна Return to thread
Alberto Romero's avatar

Hey Sheila, can you concretize your question with an example? What are you thinking exactly?

Expand full comment
Sheila Dean's avatar

Ok. For instance, we know an Meta and Anthropic LLMs grabbed user information from mobile smartphone users to train their models and there was no notice. It just raided the pantry to compensate the data sets in order to get it to work on global attributes. Who instructed the chat models to reach for these data? Did it just make a decision? Was legal ownership of the data an afterthought? Did they consider rules license? It's conspicuous that the models grabbed the information. The governing basis isn't so clear.

Expand full comment