Red Teaming

Informing AI safety and alignment discussions with adversarial testing methods currently.