“Asserts that significant AI safety improvements are achievable through private action, framing the solution as simple and within reach.”