Anthropic
From a compute build-out to long-horizon governance research and a disclosure policy, how Anthropic orders safety, research and commercial pressure after a huge raise.
Anthropic is more than a model leaderboard. A large raise pushed the race into a compute build-out, the Anthropic Institute turned governance and societal-impact work into a long-horizon programme, and a disclosure policy gives security researchers a coordinated path for flaws a model helps surface. Where the money goes, how far ahead the research commits, and who hears about a defect first all pull against each other.
