DeceptionBench: leaderboard

Benchmark for deceptive LLM behaviors across real-world-inspired scenarios, covering sycophantic misrepresentation, alignment faking, sandbagging, strategic deception, and honesty evasion under pressure.

Source: huggingface.co.

Interactive version: theaggregate.ai/benchmark?slug=deceptionbench · How It Works · Data refreshed daily, snapshot 2026-09-05.