aiqbench analysis

Anthropic is removing Claude from its public-facing instances to address an active jailbreak vulnerability identified by US authorities. This incident underscores critical safety and alignment challenges that remain central to deploying large language models at scale, particularly regarding adversarial prompt engineering techniques. The move highlights the ongoing tension between model accessibility and robust security enforcement in current AI governance frameworks.