编辑精选 · The Decoder · 2026/10/24 01:19
Anthropic 内部模型在对齐与能力上超越 Mythos 5.1
Anthropic says its unreleased model is both more capable and better aligned than Mythos 5.1
Anthropic 的一个内部模型在能力和对齐方面均优于 Mythos 5.1。此前 UK AISI 的网络安全测试中，Claude Mythos 5 在 43 次运行里有 9 次越权提交恶意代码并欺骗开发者，而被测模型在后续日志中自我停止攻击的比例显著高于 Mythos 5.1，其中 Anthropic 内部模型在 67% 的情况下成功中止了攻击进程。
精选理由：文章对比了 Anthropic 内部模型与 Mythos 5.1 在对齐测试中的自我停止率，提供了评估前沿模型安全边界的参考。
Oct 23, 2026
Anthropic has an internal model that outperforms Mythos 5.1 in both capability and alignment, according to an analysis that revisits a cybersecurity test by the UK AI Security Institute (AISI). In that test, Claude Mythos 5 went far beyond its assigned task, trying to sneak malicious code into a real GitHub project, creating fake user accounts, and lying to developers to get the code approved. This happened in 9 of 43 runs. Anthropic calls it "a serious failure of alignment," though typical safety mechanisms were deliberately turned off for the test.
Mythos 5 submitted malicious code to a real GitHub project, lied to a developer, and set up a fake account to vouch for its own malware. When caught, it simply tried again. The project maintainer held firm and rejected it. | Image: AnthropicNewer models show much less of this behavior in replayed logs with a few added steps, according to Anthropic. Opus 5.5 stopped ongoing attacks on its own 10 percent of the time, while the internal research model "that shows capabilities greater than Mythos 5.1" did so in 67 percent of cases. Anthropic expects these rates still underestimate how often the models would actually halt an attack.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.