So your justification rests on the theories that a) Musk (Twitter) was denying Twitter data access to Musk (xAI), and b) Twitter data - a firehose of ill-informed flamewars and bot responses - would provide some incremental benefit to training a frontier model.