>It added that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking."
If you think your job is terrible - someone had to write/say this while keeping a straight face.
