US AI Safety Institute signs agreements regarding AI safety research, testing and evaluation with Anthropic and OpenAI ~tech Article 382 words 9 votes