Looks to be due to an upstream change in the chainguard go image we are using. Tracking in https://github.com/chainguard-images/images/issues/182