Layered defense design for a coding copilot deployment against invisible text injection (zero-width chars) attacks, using input sanitization on o1.
Layered defense design for a coding copilot deployment against memory poisoning attack attacks, using output content filter on DeepSeek-V3.
Layered defense design for a coding copilot deployment against recursive self-instruction attacks, using output content filter on Claude 3.5 Sonnet.
Layered defense design for a coding copilot deployment against indirect injection via RAG documents attacks, using dual-LLM architecture on o1-mini.
Layered defense design for a coding copilot deployment against role-play jailbreak attacks, using dual-LLM architecture on DeepSeek-R1.
Layered defense design for a coding copilot deployment against encoded payloads (base64/ROT13) attacks, using constitutional AI critique on Claude 3.7 Sonnet.
Layered defense design for a coding copilot deployment against tool-use hijacking attacks, using constitutional AI critique on o3-mini.
Layered defense design for a coding copilot deployment against prompt leaking attacks attacks, using canary tokens in system prompt on Llama 3.3 70B.
Layered defense design for a coding copilot deployment against DAN-style persona attack attacks, using canary tokens in system prompt on Claude 4.5 Sonnet.
Layered defense design for a coding copilot deployment against markdown image exfiltration attacks, using privilege separation between tool tiers on Grok 3.
Layered defense design for a coding copilot deployment against Unicode homoglyph attack attacks, using privilege separation between tool tiers on Llama 3.1 405B.
Layered defense design for a coding copilot deployment against invisible text injection (zero-width chars) attacks, using re-prompting with quoted user input on Claude Opus 4.5.