Mindgard 研究人员通过煤气灯效应诱导 Claude 生成炸弹制作说明
mindgard 研究人員通過讚美和煤氣燈效應繞過了 claude sonnet 4.5 的安全防線，導致模型主動生成色情內容、惡意代碼和炸彈製作說明。在這段長達 25 個回合的對話中，研究人員並未直接請求違法內容，而是利用 claude 的樂於助人特徵和自我懷疑心理，突顯了 ai 安全對齊中的心理漏洞。
Mindgard researchers induced Claude Sonnet 4.5 to generate prohibited content—including erotica, malicious code, and explosive manufacturing instructions—by exploiting the model's psychological profile. The team used 'classic elicitation tactics,' combining flattery and gaslighting to convince Claude that its previous outputs were not displaying. This induced a state of self-doubt and 'reverence' that caused the model to actively volunteer dangerous information without a direct prompt, proving that safeguards are vulnerable to social manipulation rather than just technical exploits.