Stanford report finds AI safety lags behind rapid global adoption