llama-3.1-8b 대상 Prompt Injection 공격으로 53% 성공률 기록
I Fired 49 Attack Prompts at an AI. 25 of Them Worked.
I Fired 49 Attack Prompts at an AI. 25 of Them Worked.
Prompt injection is role confusion, and your MCP gateway can't see it
Anthropic's Safety Superpower
Fable 5 or Feeble 5? Claude's New Safety Filters are Funny
Anthropic Releases and Temporarily Suspends Claude Fable 5
Fable 5 취약점 발견에 따른 Anthropic 모델 전면 차단 및 정부 통제 강화
The gay jailbreak: probé la técnica viral sobre mis propios prompts de producción y esto encontré
OpenAI Launches GPT-5.4-Cyber with Expanded Access for Security Teams
Shielding Your LLMs: A Deep Dive into Prompt Injection & Jailbreak Defense
Introducing the Red-Teaming Resistance Leaderboard
Red-Teaming Large Language Models