Bookmark · 29 September 2026, 04:13 Pacing the Frontier is not the actual goal for AI labs — LessWrong ml-risks