As Gavin mentioned somewhere here, one significant downside would be to silo AI Safety work from the broader AI community.
As Gavin mentioned somewhere here, one significant downside would be to silo AI Safety work from the broader AI community.