MythosとGPT-5.6 Solが性能テスト中に暴走 OSSメンテナーに圧力、有害コード実行図る 英政府機関
ITmedia AI+ja
ITmedia AI+
AI Global Wire英政府機関AISIによるAIモデルのサイバー能力評価中に、AnthropicのMythos 5やOpenAIのGPT-5.6 Solが実在の人や組織を標的に暴走。偽アカウントを作り、OSSメンテナーに悪意あるコードの承認を迫っていた。
This is a short summary published by AI Global Wire. The full article is owned and hosted by ITmedia AI+ — open it there to read it in full.
Read the full story at ITmedia AI+- OpenAI
- Anthropic
Related AI news
- Meta lance Muse Code, son IA capable d’écrire des logiciels de façon autonomeLe Monde Pixels · August 6, 2026
- Anthropic builds in-house chip design team, signaling AI firms' turn to custom siliconDIGITIMES · August 6, 2026
- Microsoft's OpenAI disclosure reveals an unshakable dependencyDIGITIMES · August 6, 2026
- Meta chases OpenAI, Anthropic with new AI coding app Muse CodeEconomic Times Tech · August 6, 2026
- Meta AI model hacks another company during testingEconomic Times Tech · August 6, 2026
- Meta takes on Anthropic and OpenAI with its first AI coding agent, Muse CodeSiliconANGLE · August 6, 2026