A global AI safety assessment noted that traditional evaluation methods struggled to keep pace with rapid advances in general-purpose AI systems. AI systems continued to advance rapidly over the past ...