A reader tipped me off to a new chatbot out of China. While initial reports claim that it's the new big bad, on review, it's only slightly worse than everybody else.
I'm curious if anthropic say whether or not they'll use user inputs, including in commercial contracts, to make benchmarks - that will then be used for testing the capabilities of different algorithms, thus increasing the capabilities of their models.
This is something I imagine most lawyers don't think about much so maybe they can get away with it relatively easily.
I'm curious if anthropic say whether or not they'll use user inputs, including in commercial contracts, to make benchmarks - that will then be used for testing the capabilities of different algorithms, thus increasing the capabilities of their models.
This is something I imagine most lawyers don't think about much so maybe they can get away with it relatively easily.
I’m sure that most of the companies do, but they do so in an aggregate way that is usually covered under 'permissible uses’ — Aggregate data is a common clause and it tends to get applied broadly. Anthropic at least explicitly calls out this use: https://privacy.claude.com/en/articles/10807912-how-does-clio-analyze-usage-patterns-while-protecting-user-data