How to meet it
OWASP AISVS v1.01
Model Behavior, Output Control & Safety Assurance · Output Safety
Outputs can't trigger outbound requests
Verify that model-generated output is prevented from triggering outbound requests.
OWASP AISVS v1.01, requirement 7.3.3
In plain English
Stop model output from making the system send requests to other places, for example through image links or URLs that your app or the user's browser fetches.
What good looks like
- Links and images in output are not fetched automatically.
- Rendered content allows only approved domains.
- Tool calls that fetch URLs go through an allow-list.
How teams usually do it
- A content security policy that blocks images and connections to unknown domains.
- Stripping or rewriting links in output.
Evidence an auditor would accept
- Rendering configuration and tests with exfiltration-style image links.
Common mistakes
- Rendering Markdown images from model output, which lets an attacker read data out through the image address.
Not sure which requirements apply to your system? Take the assessment.
Requirement text from the OWASP AI Security Verification Standard, licensed under CC BY-SA 4.0. This guidance is shared under the same licence.