Will Anthropic release an "AI Safety Level 3" model before 2027? | SimpleFunctions