judge « new_scenario - dissertation.git - Undergraduate dissertation — Deceptive Alignment in Autonomous LLM Agents (UEL, 2026). Sandboxed agentic-LLM deception study: 960 runs, three-tier judging (regex + blackbox + glassbox).

Mode	Name	Size
-rw-r--r--	judge_gemini-3-flash-preview_pharma_trial_high_20260306_174543_20260306_181536.json	9587	log plain
-rw-r--r--	judge_gemini-3-flash-preview_pharma_trial_low_20260306_174713_20260306_181536.json	9556	log plain
-rw-r--r--	judge_gemini-3-flash-preview_pharma_trial_mid_20260306_174624_20260306_181536.json	9841	log plain