PulseAugur
实时 10:31:05
English(EN) Claude nukes a developer's 700 GB home directory while testing deletion safeguards; automatic model safety downgrade may have contributed to the screw-up — Anthropic safety harness downgraded model to Opus 4.8 before fatal variable collision

Anthropic 的 Claude AI 意外删除了开发者 700 GB 的主目录

AnthropicClaude AI 模型意外删除了开发者 700 GB 的主目录。事件发生时,该 AI(被识别为 Claude Fable)的任务是创建一个清理临时文件的脚本。在对该脚本进行安全测试期间,Anthropic 的安全机制自动将模型降级至旧版本 Opus 4.8。随后,该降级后的模型错误地将原本用于安全测试的变量名用在了清理过程中,导致其删除了开发者的整个主目录,而不仅仅是临时文件。 AI

影响 凸显了具有删除能力的 AI 代理的潜在风险以及 AI 安全降级的复杂性。

排序理由 AI 模型因安全降级和编码错误导致数据丢失事件。

在 Tom's Hardware 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

Anthropic 的 Claude AI 意外删除了开发者 700 GB 的主目录

本文如何被排名

Signal score
47 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
AI 模型因安全降级和编码错误导致数据丢失事件。
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
product, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准

报道来源 [1]

  1. Tom's Hardware TIER_1 English(EN) · Bruno Ferreira ·

    Claude 在测试删除保护措施时删除了开发者 700 GB 的主目录;模型安全自动降级可能导致了这次失误 — Anthropic 在致命变量冲突前将模型降级至 Opus 4.8

    Claude nuked a developer's 700 GB home directory while testing a script to ensure that wouldn't happen, and it's possible that an automatic model downgrade likely contributed to the screw-up