It seems that OpenAI is scraping [certificate transparency] logs

They would be far from first. Any time I create a Wildcard cert in LE I immediately see a ton of sub-domain enumeration in my DNS query logs. Just for fun I create a bunch of wildcard certs for domains I do not even use just to keep their bots busy ... not used as in parked domains. This has been going on about as long as the CT logs have existed.