OpenAI and Anthropic agree to let U.S. AI Safety Institute test and evaluate new models
The U.S. AI Safety Institute on Thursday announced it had come to a testing and evaluation agreement with OpenAI and Anthropic.