Users might paste sensitive data into a prompt. What is the risk?
Why C is the answer
Sharing sensitive data with an LLM
Pasting sensitive data into a prompt shares it with an LLM, which is a recognized risk.
In detail
- The risk is sharing sensitive data with an LLM.
- Screen inputs, and train staff.
Use case
An illustrative scenario, not a real incident.
A user pastes customer records into a prompt, which shares them with the model.
See it in different ways
flowchart LR A["Prompt"] --> B["Model Armor"] B --> C["Gemini"] C --> D["Safety filters"] D --> E["Response"]
sequenceDiagram participant U as User participant A as Model Armor participant G as Gemini participant F as Output filter U->>A: Prompt A->>G: Screened prompt G->>F: Response F-->>U: Regex and safety checks
mindmap
root(("Risks"))
n1["Data"]
n2["Sensitive data in prompts"]
n3["Data exfiltration"]
n4["Attacks"]
n5["Malicious prompting"]Key point: Prompts can leak sensitive data.
Why the other options are wrong
- A. A lower accelerator quota
- That is not a privacy risk.
- B. A smaller Feature Store
- That is not a privacy risk.
- D. A slower training job
- That is not a privacy risk.