Benign-trained probes miss tool-action errors under adversarial conditions
Scalar uncertainty beat the probe on AUROC and average precision for Qwen2.5-14B-Instruct under system override.
42
AI model
Qwen2.5-14B-Instruct
- Mentions · 7d1
- Mentions · 30d1
- All mentions2
- First seenSep 27, 2026
- Last mentioned2d ago
Timeline
2 articles · newest firstSee it, Say it, Sorted: Mechanistic Diagnosis and Parameter-Space Mitigation of Emergent Misalignment in LLMs
A geometric mitigation framework cuts emergent misalignment on Qwen2.5-14B-IT by up to 80%.
62
Appears with
- AgentDojoproduct · 1
- Alibabaorganization · 1
- Qwen2.5-7B-Instructai model · 1
- GLM-4.6V-Flashai model · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.