The incident has been closed on the Heroku status page, and autoscaling is up and running.

We're starting to see successful autoscales for Heroku now, as their API is recovering.

Heroku has just opened an incident on their status page: https://status.heroku.com/incidents/2912. We'll continue to monitor.

We're seeing an increase in errors from Heroku when trying to talk to their API and trigger autoscaling, started around 12:05 UTC.