🧰 gpt-4o-mini

The items that name it 4

• Threat-Preserving Representation Sensitivity in Agent-Security Benchmarks
  paper2026-10-02
• LLMjacking evolved: Attackers are using stolen AI compute to build offensive agentic tools | Sysdig
  writeup2026-06-17It is written here as gpt-4o-mini (OpenAI).

  OpenAI

• PISmith: Reinforcement Learning-based Red Teaming for Prompt Injection Defenses
  paperIt is written here as GPT-4o-mini (closed-source LLM attacked in agentic settings).

  closed-source LLM attacked in agentic settings

• Sockpuppeting: How a Single Line Can Bypass LLM Safety Guardrails | TrendAI (US)
  writeup

  Most resistant model tested, ASR of 0.5%