The White House AI accord outlines sensible safeguards, including independent evaluators and board-level oversight, but does not compel companies to adopt them. Until the cost of safety failures outweighs the payoff of racing ahead, a pledge built on “believe” and “should” won’t slow anyone down.