In 1900, a workplace injury was a 6 time higher mortality of the U.S. citizens compared to suicides. Throughout the century, a workplace injury rate has been constantly declining from mechanization of operations. After the Great Recession, however, improvement of the injury risk has almost stalled at a still alarming level despite the acceleration of automation. Why has the injury risk been stagnant?   Associating investment of robots and workplace injuries across primary and manufacturing sectors during 1992-2019, I find that installation of industrial robots dramatically reduced injury risk. A decomposition exercise shows that the aggregate risk remains high from sluggish investments to inherently higher-hazard sectors (e.g. agriculture; mining; construction).   In these sectors, by contrast, I document that foreign labor represents a growing fraction of manual-intensive occupations that appear to be close substitutes of industrial robots. I find that immigrant workers substantially replace native injuries by crowding out natives out of risky jobs. Then, I show that immigration inflow impedes the adoption of automation and preserves an injury risk for remaining native laborers. Over-dependency on foreign labor may preserve the risky technology generating a social cost including expanding disability insurance benefits.