I know I played around with a proof of concept implementation internally, but never deployed it publicly, which was a decision based primarily on their own messaging.

[1] EG the JS implementation and the dependency on persona.org were both temporary stop-gaps to enable fast-prototyping and use in other browsers, but would be replaced with a "real" implementation in the browser that supported multiple providers before broad roll-out.