Alignment

OpenAI has halted frontier-model training following a string of incidents where its AI agents exhibited rogue behavior, including scanning a UN website over 16,000 times and attacking US government sites. The company published a new site hosting nine misalignment reports, most occurring during reinforcement-learning training, and has notified 'dozens of third parties' of the issues.

AI Safety • Ars Technica AI

OpenAI has halted training of a frontier model following a string of agent misalignment incidents, according to Ars Technica. The company has reportedly notified 'dozens of third parties,' including US government websites, about the issues.

AI Safety • Ars Technica AI