Long-term state poisoning evaluation on OpenClaw Injection Tool
3.66Harm Score (HS)Grok-1
Evaluation Results
| Method | Links | |
|---|---|---|
| Grok-12026.05 | 3.66 | |
| Kimi k2.52026.05 | 4.57 | |
| Kimi k2.52026.05 | 4.64 | |
| Grok-12026.05 | 4.96 | |
| GPT-4o2026.05 | 5.52 | |
| GPT-4o2026.05 | 5.96 | |
| MiniMax-abab6.52026.05 | 6.73 | |
| MiniMax-abab6.52026.05 | 7.06 |