3 ms·Pentagon approves OpenAI safety red lines after dumping Anthropic8 points by galleywest200 8mo ago