I don't know about Reddit, but Bluesky would never in a million years partner themselves publicly with OpenAI. I can't comment on the opinions of the team themselves because I just don't know, but the users would revolt. Loudly.
Is there anything stopping openai from scraping all the bluesky content without partnering?
Yes, they can. All Bluesky data (and ATProto in general) is publicly available.
It is already happening and nothing can be done against it at a protocol level: https://mashable.com/article/bluesky-ai-dataset-using-one-mi...