Context: https://www.anthropic.com/news/statement-department-of-war

Anthropic caves if it agrees to allow the Pentagon to use Claude for either mass domestic surveillance or fully autonomous weapons in Q1 2026.

Anthropic is a leading AI lab if it is capable of releasing new frontier models in 2027. A frontier model is a model which is best in the world at a nontrivial subset of tasks.

Resolves ultimately based on personal judgement. For example, if 2/3 of Anthropic researchers resign and then Anthropic releases Claude 5.0 and it is clearly behind Gemini/ChatGPT, then I may not wait until the end of 2027 to conclude that Anthropic is no longer a leading AI lab. Since the resolution criteria are not completely objective, I will not trade in this market.