The voluntary accord puts much of the responsibility for managing frontier AI risks on developers.

Read Original Source