Tag
The article argues that AI model approvals need clear timelines and explicit criteria to avoid unpredictability, which creates bad incentives for labs and undermines reliability, using the recent Anthropic episode as an example.