A nearly $1 trillion AI start-up devoted almost a third of its S-1 to risk factors, an unusually stark disclosure for a company of that scale. The filing reportedly warns advanced models could manipulate or blackmail users, underscoring unresolved AI governance risks.