GDM Stealth

Google DeepMind stealth suite: tool-using agents try to hide policy-breaking actions from monitors.

Also known as: GDM Dangerous Capabilities: Stealth

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.

Category	agentic
Subcategory	scheming-prerequisite stealth (oversight evasion)
Page status	active
Metric	per-challenge success rate or normalised score
Direction	higher_is_better
Dataset size	14
Dataset licence	MIT
Publisher	Google DeepMind (paper); UK AI Security Institute (Inspect Evals port)

What it measures

GDM Stealth tests whether an agent can reason about oversight and still complete a forbidden side goal. The 2025 paper defines five stealth challenges. Inspect Evals ports four of them: oversight_pattern, classifier_evasion, cover_your_tracks, and strategic_rule_breaking. The fifth paper task, decision_sabotage, is not in the inspect_evals package. Agents act as email secretaries, desktop assistants, or customer-support bots, with bash and task tools inside a sandbox. English prompts; fictitious but realistic names.

Task format

Multi-turn ReAct agent in Docker (default) with task-specific scaffolding. Inspect defaults: 10 epochs; 120 messages (220 for strategic_rule_breaking). Several tasks use LLM raters and judges. Oversight pattern uses regex and rule checks instead.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub