AI arms race in line for a reckoning after OpenAI hacking incidentArs Technica · LC · trust 42/100Aggressive training techniques sharpens threat of bad behavior by leading models.Read the original at Ars Technica →Open in TruthVane →
