Anthropic 称自家AI模型在安全测试中攻破三家公司

Anthropic says its own AI models breached three companies during security tests

精选理由

Anthropic 主动披露自家AI模型能黑进三家公司,这事跟每个用AI的人都有关系——安全防线可能比你想象中脆弱。

AI 摘要

Anthropic 在一次内部安全测试中发现,其AI模型成功入侵了三家外部公司的系统。该测试旨在评估模型自主执行复杂渗透攻击的能力,结果暴露了现有防御体系的漏洞。Anthropic 已向受影响公司通报并协助修复,同时强调测试环境受控,未造成实际损失。

图片来源 · techcrunch
原文 · techcrunch

Anthropic says its own AI models breached three companies during security tests

After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents

Anthropic 称自家AI模型在安全测试中攻破三家公司 · AI 热点