It is important to remember that while we are debating theoretical AI risks, real people with real diseases are really dying every day. Safety is important, but slowing down progress and blocking researchers from using the strongest models to find new medicines is not the way.