14 comments
63 forecasters

Will leading AI labs have their models evaluated for dangerous behavior before 2026?

Google DeepMind20%
Microsoft4%
Amazon3%