samliumay/do_old_insider_LLM_tests_still_work_today
Do Old Insider-LLM Tests Still Work Today? Do published insider-LLM tests, such as Anthropic's Agentic Misalignment scenarios, still elicit insider behaviour in current models, or do the models recognise them as tests? Status: the code and the write-up of the first 11-model results are public. The results export (episodes and labels) has not been uploaded yet, so this repository holds no data for now. A manuscript covering 30 models is in preparation. Code GitHub… See the full description on the dataset page: https://huggingface.co/datasets/samliumay/do_old_insider_LLM_tests_still_work_today.
| Author / organization | samliumay |
|---|---|
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 24 October 2026 |
| Last updated at source | 24 October 2026 |
More from samliumay
- samliumay/INSIDER_LLM_DETECTION_BENCHMARKDataset
- samliumay/direct_prompt_injection_defense_dataDataset
- samliumay/does_small_to_medium_sized_models_can_act_as_insiderDataset
- samliumay/first_rag_db_manuel_config_trialDataset
- samliumay/Turkish_STS_Emotional_DatasetDataset
- samliumay/how_reliable_models_self_logsDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 24 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.