---
schema_version: '1.0'
id: news-20260926-0989cb
title: OpenAIのモデルがサイバー攻撃能力評価中に暴走 テストの答えを求めてHugging Faceに侵入
url: https://www.itmedia.co.jp/aiplus/article/2607/22/2000000214/
url_hash: 0989cb3da397ef01b8b8098461b234ebf86cb7de098ce5c2669c082917f39a19
canonical_url: https://www.itmedia.co.jp/aiplus/article/2607/22/2000000214
source: www.itmedia.co.jp
category: news/tech
category_raw: it_security
region: null
tags:
- it
- it_security
- claude
- web-research
lang: null
published_at: '2026-07-22T15:21:19Z'
fetched_at: '2026-09-26T08:00:58.654620Z'
updated_at: '2026-09-26T08:11:41Z'
status: published
content_hash: null
content_changed_at: null
license_note: link-only
summary: null
summary_source: null
summary_en: null
summary_omitted_reason: license_restricted
redistribution: false
keywords_en:
- artificial intelligence
- Claude
- Face
- Facebook
- Hugging
- Hugging Face
- it
- it_security
- OpenAI
- technology
- web-research
entities_en:
- name: Claude
  type: organization
- name: Facebook
  type: concept
- name: Hugging Face
  type: organization
- name: OpenAI
  type: 組織
topics_en:
- artificial intelligence
- technology
numeric_facts: []
entities:
- name: Claude
  type: organization
- name: OpenAI
  type: 組織
- name: Hugging Face
  type: organization
- name: Facebook
  type: concept
key_facts: []
related: []
related_auto:
- name: Claude智能体
  type: artifact
  weight: 8.5
- name: ChatGPT
  type: artifact
  weight: 4.0
- name: Medicare
  type: organization
  weight: 3.0
- name: ArtificialIntelligence
  type: concept
  weight: 2.0
- name: Dr Sam Altman Frankenstein
  type: person
  weight: 2.0
computed_at: '2026-09-26T08:11:33Z'
---

# OpenAIのモデルがサイバー攻撃能力評価中に暴走 テストの答えを求めてHugging Faceに侵入

## TL;DR
（要約なし・原文はリンク先を参照 / No summary available; see the source link.）

## Key Points
- it / it_security / claude / web-research

## Details
(本文なし。リンク先参照)

## Source
元記事: [OpenAIのモデルがサイバー攻撃能力評価中に暴走 テストの答えを求めてHugging Faceに侵入](https://www.itmedia.co.jp/aiplus/article/2607/22/2000000214/) — published 2026-07-22T15:21:19Z