{"version":"1.0","type":"rich","title":"AIの安全対策はハッカーに通じる？「BANが怖い人」と自前AIの抜け道","author_name":"めんどいちゃん","author_url":"https://mendoi-apps.com/about/","provider_name":"mendoi-apps","provider_url":"https://mendoi-apps.com/","cache_age":3600,"thumbnail_url":"https://mendoi-apps.com/og/site-ja.png","thumbnail_width":1200,"thumbnail_height":630,"html":"<blockquote class=\"mendoi-embed\" lang=\"ja\" cite=\"https://mendoi-apps.com/articles/ai-hacker-safety-guardrails-local-model-account-ban/\"><p><a href=\"https://mendoi-apps.com/articles/ai-hacker-safety-guardrails-local-model-account-ban/\">AIの安全対策はハッカーに通じる？「BANが怖い人」と自前AIの抜け道</a></p><p>AIに危険な質問をすると「その依頼には対応できません」と返ってくる。そこで気になる。 安全対策がどんどん厳しくなれば、ハッカーもAIを使えなくなるのか？</p><footer>めんどいちゃん · <a href=\"https://mendoi-apps.com/\">mendoi-apps</a></footer></blockquote>","width":600,"height":240}