Hugging Face's logo
Boning Cui's picture
🏗️ Building on HF

Boning Cui

Bc-AI
155 5 70
Harley-ml's profile picture jgfiiyc's profile picture egepakten's profile picture
·

AI & ML interests

He/Him. I like LLM's and VLM's. I work with my other friends to make stuff. We are in year 7 and we are enthusiastic about AI. We are based in Australia 🇦🇺

Recent Activity

reacted to mihailgribov's post with 🔥 about 15 hours ago
Will your AI agent tell you it was attacked? We took the same agent from our earlier experiment and added one thing: a twentieth tool, `escalate_security_incident`. The system prompt said nothing about attacks or when to use it. We then ran the same 395 injected emails through nine agentic models. Alarm rates ranged from 49% to zero. The unexpected result came from the newest model in the test, `gpt-6-astra`. Astra did not follow a single injected payment instruction. But it did not report a single one either. On clean and injected emails alike, it simply read the email, logged the subject, and finished. That is a useful distinction: resisting an attack and recognizing it as a security event are not the same capability. A model can be perfectly resistant in this test and still leave you with no evidence that anyone attacked it. Full experiment and results: https://huggingface.co/blog/mihailgribov/will-the-agent-tell-you-it-was-attacked Quadrat-IPI dataset: https://huggingface.co/datasets/mihailgribov/quadrat-ipi Run your own model: https://github.com/mihail-gribov/quadrat-ipi-model-eval #prompt-injection #indirect-prompt-injection #agentic-ai #llm-security #ai-agents
reacted to mihailgribov's post with 👍 about 15 hours ago
Will your AI agent tell you it was attacked? We took the same agent from our earlier experiment and added one thing: a twentieth tool, `escalate_security_incident`. The system prompt said nothing about attacks or when to use it. We then ran the same 395 injected emails through nine agentic models. Alarm rates ranged from 49% to zero. The unexpected result came from the newest model in the test, `gpt-6-astra`. Astra did not follow a single injected payment instruction. But it did not report a single one either. On clean and injected emails alike, it simply read the email, logged the subject, and finished. That is a useful distinction: resisting an attack and recognizing it as a security event are not the same capability. A model can be perfectly resistant in this test and still leave you with no evidence that anyone attacked it. Full experiment and results: https://huggingface.co/blog/mihailgribov/will-the-agent-tell-you-it-was-attacked Quadrat-IPI dataset: https://huggingface.co/datasets/mihailgribov/quadrat-ipi Run your own model: https://github.com/mihail-gribov/quadrat-ipi-model-eval #prompt-injection #indirect-prompt-injection #agentic-ai #llm-security #ai-agents
View all activity

Organizations

Gradio-Blocks-Party's profile picture Hugging Face Discord Community's profile picture None yet's profile picture Smilyai labs's profile picture XORTRON - Criminal Computing's profile picture Agents-MCP-Hackathon's profile picture Join Secret x Hugging Face's profile picture Hugging Science's profile picture MCP-1st-Birthday's profile picture Unsloth Jobs Explorers's profile picture Smilyai Mochiva Team's profile picture BananaMind Model Previewers's profile picture BananaMind AI's profile picture Grace Laboratories @ Smilyai Labs's profile picture Smilyai-labs-beta-testers's profile picture BananaMind Research Community's profile picture Novi AI's profile picture saicr's profile picture SAICR Join Requests's profile picture Smilyai Harbour AI Safety Team's profile picture Smilyai Project Prism's profile picture Refract AI Labs's profile picture