OpenAI 和 Anthropic 竞相部署高风险的网络安全人工智能

OpenAI 将推出新的网络安全模型来挑战 Anthropic 的克劳德神话，重点关注先进的防御和黑客能力。

Kerem Gülen · April 10, 2026, 13:42 ·1 min read

据消息人士透露，OpenAI 将推出一种网络安全模型，旨在与 Anthropic 的 Claude Mythos 竞争。新模型将通过 OpenAI 的 Trusted Access for Cyber 试点计划向选定的一组企业提供。

该举措于二月份首次宣布，旨在增强基线保障，同时测试防御加速的可信访问。 OpenAI 承诺向参与试点计划的企业提供 1000 万美元的 API 积分。

Anthropic 最近宣布计划向一小部分公司发布 Claude Mythos Preview。目的是帮助组织识别和纠正软件漏洞，以回应人们对关键基础设施中人工智能滥用的日益担忧。

政府官员对人工智能工具破坏金融网络和电网等重要系统的可能性发出了警告。全球 IT 和网络安全提供商 RFA 表示，人工智能正在加剧网络安全风险，特别是对于私募股权公司而言。

<小时/>

精选图片来源

Figma 和 Anthropic 合作开发代码到画布 AI 功能

Written by

Kerem from Turkey has an insatiable curiosity for the latest advancements in tech gadgets and a knack for innovative thinking. With 3 years of experience in editorship and a childhood dream of becoming a journalist, Kerem has always been curious about the latest tech gadgets and is constantly seeking new ways to create. As a Master's student in Strategic Communications, Kerem is eager to learn more about the ever-evolving world of technology. His primary focuses are artificial intelligence and digital inclusion, and he delves into the most current and accurate information on these topics.

View all posts →