OpenAI rejects researchers’ claim they were fired over safety warnings, says they committed ‘breach of trust’