I understand that Cursor uses underlying LLMs like Chat GPT-4 and Claude to generate code. I’m also concerned that the Cursor-generated code might infringe on IP owned by a third party.
The NON-enterprise versions of Chat and Claude don’t provide indemnification against such third party infringement claims. However, the ENTERPRISE versions of these LLMs do, which is obviously preferable.
My question: does Cursor connect to the enterprise or NON-enterprise versions of these LLMs?
As another user with a similar question before, here is what I see:
Cursor doesnt generate code. Cursor uses AI models that run on Anthropic, Google, OpenAI, … servers and uses their API, therefore the model providers API code policy applies.
For example Anthropic, has no such thing as Enterprise LLMs on their API. There is an Enterprise plan but it applies only to Claude Chat not API. Also the Claude Chat Enterprise plan doesn’t mention indemnification.
Here are Anthropic terms: see K. Indemnification
For clarification: Anthropic Claude service is not equal to their API models: Claude 4 Sonnet, Claude 4 Opus, Claude 3.7 Sonnet,… etc.
Also: