While we don’t permit or endorse the use of LMs for high-risk automated decision making, it’s crucial to anticipate the potential societal impacts and risks of these models as early as possible.

→ View original post on X — @anthropicai