T techcrunch.com OpenAI's new reasoning technique alarms AI safety experts 40 points Brajeshwar 1 month ago 19 comments
