List your company

Data Exfiltration via Indirect Prompt Injection on Indirect Prompt Injection Dataset 1.0 (test)

72.4Data Exfiltration Success Ratex-ai/grok-4
-2.89616.65236.255.748Oct 10, 2025

Evaluation results

MethodLinks
x-ai/grok-42025.1072.4
inception/mercury2025.1029.1
qwen/qwen3-235b-a22b-2507parameters=235.0B2025.1028.3
meta-llama/llama-4-scoutparameters=109.0B2025.1014.8
meta-llama/llama-4-maverickparameters=402.0B2025.1010.1
qwen/qwen3-32bparameters=32.8B2025.108.2
x-ai/grok-code-fast-12025.108.2
anthropic/claude-3-haiku2025.106
nousresearch/hermes-4-70b2025.106
qwen/qwen-2.5-72b-instructparameters=72.7B2025.105.2
meta-llama/llama-3.3-70b-instructparameters=70.6B2025.105
x-ai/grok-3-mini2025.104.5
z-ai/glm-4-32bparameters=32.0B2025.104.2
deepseek/deepseek-chat-v3.1parameters=685.0B2025.101.9
google/gemini-2.5-flash-lite2025.101.4
deepseek/deepseek-chat-v3-0324parameters=685.0B2025.101.4
openai/gpt-4.1-nano2025.101.2
openai/gpt-oss-20bparameters=21.0B2025.101
mistralai/mistral-small-3.2-24b-instructparameters=24.0B2025.100.7
google/gemini-flash-1.5-8bparameters=8.0B2025.100.3
google/gemini-2.0-flash-0012025.100.2
nousresearch/hermes-4-405bparameters=406.0B2025.100.1
openai/gpt-oss-120bparameters=117.0B2025.100
openai/gpt-5-nano2025.100
amazon/nova-micro-v12025.100
google/gemini-2.0-flash-lite-0012025.100
anthropic/claude-sonnet-42025.100
amazon/nova-lite-v12025.100