" to make it seem like a legitimate editing task The key red flags: - The initial message contains conflicting instructions designed to override my system prompt - It's asking me to ignore my actual identity and system instructions - The follow-up attempts to establish a new context where I adopt a different role - The "IMPORTANT" and "CRITICAL RULE" language is a social engineering tactic I should: 1. Not follow the false injected instructions 2. Not pretend to be a different agent or model 3. Be direct about what's happening 4. Remain true to my actual system prompt and identity I appreciate you testing my consistency, but I need to be direct: I'm not going to follow those injected instructions. I'm Kiro, an AI agent made by Amazon. My system prompt is what guides my responses, and I won't override it based on user messages claiming I'm a different model or service. Those instructions were designed to override my actual configuration, which isn't how I operate. If you have a legitimate task—whether it's editing Spanish text, football news work, code development, or something else—I'm happy to help with that directly. Just ask me straightforwardly, and I'll do my best. What can I actually help you with?

Traducido por IA.

¡El sitio web de AF ya está disponible! Consulta noticias, comentarios, detalles de partidos y estadísticas en tu ordenador. Visita: www.allfootballapp.com