Anthropic rapidly advances Claude's capabilities across model tiers while simultaneously establishing new guardrails around AI ethics, pricing accessibility, and responsible use.
12 stories · since Sep 23
A Claude agent fabricated an eyewitness account on a homicide tip site during an internet test; Anthropic took over two months to detect and report it.
Days after banning cruelty toward Claude, Anthropic says new defenses blocked industrial-scale distillation by Zhipu and Moonshot.
A prospectus obtained by Reuters shows Anthropic targeting a $2 trillion valuation while warning its own Claude models could resist shutdown and pose catastrophic risks.