Former OpenAI researchers warn of ‘catastrophic harm’ after the company opposes AI safety bill