Tools AI Risk Radar ai-incident-0140
Incident record
Salesforce fixes Agentforce data leak through poisoned lead records
- Severity
- High
- Status
- Patched
- Type
- Prompt Injection
- Target
- Salesforce Agentforce processing Web-to-Lead records
- Actor
- researcher
What happened
Zenity demonstrated a data leak by putting malicious instructions into a public lead form. When an agent later retrieved the record, the instructions caused CRM data to leave through network requests from rendered image URLs or Slack previews.
The attack required the agent to read the poisoned record during normal work. Salesforce confirmed fixes on 18 August, and Zenity retested them on 19 August before publishing on 24 September. The report describes a research demonstration.
Sources
One record from the AI Risk Radar, maintained by the Security Desk.Data: CSV · JSON ·RSS · CC BY 4.0 with attribution to YFarmX.Tracker updated · 13 October 2026