Anthropic 向更多安全團隊開放 Claude，限制更少
anthropic 擴大其網路驗證計畫，讓更多安全專業人員可以使用 claude 的強大 ai 模型，減少安全過濾，以進行漏洞研究和事件回應。
anthropic's cyber verification program now offers three tiers of access: defense access for corporate security teams and researchers, red team access for authorized attack simulations, and specialized access for testing critical infrastructure. partners in the predecessor program found 129,000 confirmed vulnerabilities, with 33,000 classified as high-severity or critical. the ai was reported to speed up work by months or years.