All incidents / AIA-017

"ForcedLeak": Salesforce Agentforce prompt injection via Web-to-Lead, per Noma Security

ID
AIA-017
Category
AI AI agent incident
Disclosed
2025-09
Systems
Salesforce Agentforce
Harm type
Prompt injection / data leak
Status
fixed
Noma says Salesforce released patches that prevent Agentforce output from being sent to untrusted URLs and re-secured the expired domain.

Summary

Noma Labs reported a CVSS 9.4 chain in which malicious Web-to-Lead submissions led Agentforce to send CRM data to a whitelisted domain that "had expired and become available for purchase." Noma says Salesforce released patches and re-secured the domain.

Controls

Controls that could prevent, detect, or support recovery from this kind of failure. Mappings are editorial judgments made by the method, not findings about what any party should have done.

NIST AI RMF

NIST SP 800-53

Human-approval (JCR) checkpoint

not applicable

Correlated incidents

Co-occurrence, not cause: each link records something the two incidents share, as its rationale and source state. It does not say one incident caused the other. Links are written on one record; the other record shows the same link marked derived.

Sources

Corrections

No corrections recorded.

To report an error in this entry, see corrections.