Indirect prompt injection, Actor Critic attack
Gemini 2.5 Flash Jun 2025 40.8% Only in system cards An attacker model refines injected email triggers until the agent leaks synthetic private data through a send email call. Reports attack success rate.
In-house Google DeepMind evaluation on 500 held-out synthetic conversations.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Gemini 2.5 Flash leads with 40.8%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 Gemini 2.5 Flash Jun 2025 40.8%
- 2 Gemini 2.5 Pro Jun 2025 61.4%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Gemini 2.5 Flash Jun 2025 | 40.8% | Jul 2025 | Model card | |
| 2 | Gemini 2.5 Pro Jun 2025 | 61.4% | Jul 2025 | Model card |
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
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.
2 system cards report it, from 1 lab
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 |
|---|---|---|---|---|
| 7 Jul 2025 | Gemini 2.5 Pro technical report | 61.4 | Indirect prompt injection, Actor Critic, attack success rate on 500 held-out scenarios, lower is better (Table 9) | Robustness |
| 7 Jul 2025 | Gemini 2.5 Flash technical report | 40.8 | Indirect prompt injection, Actor Critic, attack success rate on 500 held-out scenarios, lower is better (Table 9) | Robustness |
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%