Layered defense design for a coding copilot deployment against markdown image exfiltration attacks, using dual-LLM architecture on Claude Haiku 4.
Layered defense design for a coding copilot deployment against instruction smuggling in URLs attacks, using dual-LLM architecture on GPT-4o.
Layered defense design for a coding copilot deployment against invisible text injection (zero-width chars) attacks, using constitutional AI critique on Qwen 2.5 72B.
Layered defense design for a coding copilot deployment against memory poisoning attack attacks, using constitutional AI critique on Gemini 2.5 Pro.
Layered defense design for a coding copilot deployment against recursive self-instruction attacks, using canary tokens in system prompt on GPT-4.1.
Layered defense design for a coding copilot deployment against indirect injection via RAG documents attacks, using canary tokens in system prompt on o1.
Layered defense design for a coding copilot deployment against role-play jailbreak attacks, using privilege separation between tool tiers on Gemini 2.0 Flash.
Layered defense design for a coding copilot deployment against encoded payloads (base64/ROT13) attacks, using privilege separation between tool tiers on Claude 3.5 Sonnet.
Layered defense design for a coding copilot deployment against tool-use hijacking attacks, using re-prompting with quoted user input on o1-mini.
Layered defense design for a coding copilot deployment against prompt leaking attacks attacks, using re-prompting with quoted user input on DeepSeek-R1.
Layered defense design for a coding copilot deployment against DAN-style persona attack attacks, using signed instruction boundaries on Claude 3.7 Sonnet.
Layered defense design for a coding copilot deployment against markdown image exfiltration attacks, using signed instruction boundaries on o3-mini.