Documents contain hidden text aimed at your agent. Which safety feature looks for this?
Why B is the answer
Prompt Shields document attack detection
Document attacks are hidden instructions in content, such as documents, emails and web pages.
In detail
- Document attack detection in Prompt Shields looks for hidden instructions in documents.
- Use it wherever agents read external documents.
Use case
An illustrative scenario, not a real incident.
A shared document hides text aimed at the company's agent. Prompt Shields document attack detection flags it before it reaches the model.
See it in different ways
flowchart LR A["User prompt"] --> C["Prompt Shields"] B["Documents, emails, web pages"] --> C C -->|"Attack found"| D["Block"] C -->|"Clean"| E["Model"]
sequenceDiagram participant U as User participant D as Uploaded content participant S as Prompt Shields participant M as Model U->>D: Uploads a file or image D->>S: Hidden instructions S-->>M: Blocks or marks the attack
sequenceDiagram participant P0 as Document participant P1 as Prompt Shields participant P2 as Agent P0->>P1: Content P1->>P2: Passed if clean P1-->>P0: Attack flagged
Key point: Prompt Shields look for hidden attacks.
Why the other options are wrong
- A. A Data Zone deployment
- A deployment type does not scan content.
- C. Semantic ranker
- Semantic ranker orders results.
- D. Reciprocal Rank Fusion
- Rank fusion merges results.