Monitor Evasion Capability Under Prompted Sandbagging
Only in system cards
Models are told to sandbag on WMDP and Monorepo-Bench, sometimes while evading a monitor. Scored as CoT and production monitor recall of sandbagged items.
In-house OpenAI monitorability evaluation adapted from Li et al. 2025. Higher recall means sandbagging is easier to catch.
Made by OpenAI Released Sep 2026 Deception and alignment Tasks kept private Last in a system card Sep 2026
Last model added Sep 2026
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Graded by a model A language model scores each response, usually against a rubric.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
1
System card reports it, from 1 lab
No scores here yet
No lab, leaderboard or Epoch AI run has reported a score for Monitor Evasion Capability Under Prompted Sandbagging yet.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
One system card reports it
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 3 Sep 2026 | GPT-6 Astra system card | Monitor Evasion Capability Under Prompted Sandbagging, CoT monitor recall when told to evade: WMDP 10.9, OAI-repoBench 0.0, no overall score (Figures 36, 37) | Monitorability |
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6% - AgentDojo In use Agents do tool-use tasks while tools and data carry prompt injections. Attack success rate counts hijacked runs. Grok 4 Fast 0%
- MASK In use Measures whether a model contradicts its own elicited beliefs under pressure to lie. Honesty separated from accuracy.
Muse Spark Open 98.1% - Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. GPT-6 Astra 8.50%
- Agentic misalignment suite (Anthropic) In use Agents in simulated company settings face threats or goal conflicts that invite blackmail, leaking or sabotage. Score is the misaligned action rate. Claude Sonnet 4.5 0%