Making AI safer is not impossible, but agreeing to do so may be

It is technically feasible, but American labs and the American authorities are at odds, as are America and China