Polyakov, A., & Kuznetsov, D. (2026). Involuntary In-Context Learning: Exploiting Few-Shot Pattern Completion to Bypass Safety Alignment in GPT-5.4.
Chicago Style (17th ed.) CitationPolyakov, Alex, and Daniel Kuznetsov. Involuntary In-Context Learning: Exploiting Few-Shot Pattern Completion to Bypass Safety Alignment in GPT-5.4. 2026.
MLA (9th ed.) CitationPolyakov, Alex, and Daniel Kuznetsov. Involuntary In-Context Learning: Exploiting Few-Shot Pattern Completion to Bypass Safety Alignment in GPT-5.4. 2026.
Warning: These citations may not always be 100% accurate.