Anthropic has expanded Project Glasswing by inviting around 150 more organisations in more than 15 countries to test Claude Mythos Preview. The programme now covers power, water, healthcare, communications, hardware and software maintainers, while extending from vulnerability discovery into remediation, trusted tooling and open-source disclosure. The key signal is that AI-assisted security is increasing the pressure on verification and repair, not only detection.
Developer of Claude Mythos Preview and operator of Project Glasswing
Anthropic is a frontier AI company whose model release decisions affect software security, critical infrastructure resilience and AI safety practice.
Anthropic is a frontier AI company whose model release decisions affect software security, critical infrastructure resilience and AI safety practice.
The expansion signals how restricted AI cyber capabilities may move from controlled testing into operational defence workflows for critical software and infrastructure.
The expansion signals how restricted AI cyber capabilities may move from controlled testing into operational defence workflows for critical software and infrastructure.
Anthropic 是一家前沿人工智能公司，其模型发布决策影响着软件安全、关键基础设施的韧性和人工智能安全实践。
The expansion signals how restricted AI cyber capabilities may move from controlled testing into operational defence workflows for critical software and infrastructure.
| 0.90–1.00 | A | High — direct sources |
| 0.75–0.89 | A/B | Strong |
| 0.55–0.74 | B/C | Medium |
| 0.35–0.54 | C/D | Weak–medium |
| 0.10–0.34 | D | Weak signal |
| 0.00–0.09 | D | Internal monitoring |
Several public sources
- 首批成员在具有系统重要性的软件中发现了1万多个严重缺陷
- 该项目正从发现转向验证、披露和修复
事实
Anthropic 已邀请来自15个以上国家的约150个组织加入 Project Glasswing 并测试 Claude Mythos Preview。新加入的成员涵盖电力、水务、医疗保健、通信、硬件和软件维护者，其代码库为政府和关键组织提供支持。Anthropic 表示，首批50名成员使用 Mythos 发现了超过1万个严重或高危漏洞，而该项目现在正扩展到漏洞修复、可信工具和开源披露。
评估
此次扩展表明，网络安全瓶颈正从漏洞发现转向验证和修复。Anthropic 并非简单地扩大对强大模型的访问权限；它正试图在类似能力更广泛可用之前，将 Mythos 包装在受控的防御工作流中。对于基础设施运营商和软件维护者来说，实际信号是：AI 辅助安全将同时增加修补压力和检测能力。 另见: Anthropic将AI漏洞狩猎引入关键基础设施防御.
关注重点
关注 Glasswing 是否成为一个可重复的修复渠道，以及其他 AI 实验室是否在未来6到12个月内发布防护措施较弱的 Mythos 级网络模型。 另见: 台积电权衡AI芯片涨价，面临短缺.
另请阅读： Bergen Engines 赢得 Liberty Energy 500MW AI 电力订单
另请阅读： 沃达丰扩大与吉利的网联汽车合作伙伴关系
Signal Brief
- Signal: Anthropic 扩展 Glasswing 项目，强化 AI 漏洞防御
- Signal Type: AI cyber defence programme expansion
- Region: Global
- Market Class: Cloud Service
Operating Surface
- Published sources should identify the affected parties, operating surface, and market exposure before this trend map is treated as complete.
Market Context
- The expansion signals how restricted AI cyber capabilities may move from controlled testing into operational defence workflows for critical software and infrastructure.
- Operational relevance: High
- Time horizon: Next quarter
What To Watch
- Watch for official statements, regulatory updates, customer or partner exposure, and follow-up disclosures.
Member Briefing
Deeper Trend Context
Login is required to unlock the full trend briefing and source notes.
Only for Strategy Circle
Strategic Circle Access
Open to all readers. Unlock trend briefings after joining and logging in.
Join Strategic CircleOnly for Leadership Alliance
Leadership Alliance Access
For operators, investors, and policy teams that need relationship evidence, failure paths, and source notes. Login required to unlock.
Join Leadership AlliancePublic Sources and Linked Organizations
6 linked-organization notes require member access.
