Skip to main content
Because AI agent widgets support anonymous users, spam cannot be permanently blocked at the individual user level. This article explains how to stop an active spam conversation and reduce repeated spam — not how to globally ban a user.

​
Why spam happens

AI agent widgets are:
  • Publicly accessible
  • Always online
  • Easy to target at scale
This makes them attractive to automated scripts that try to:
  • Probe for vulnerabilities
  • Inject links or prompts
  • Test downstream integrations
  • Flood inboxes with noise.
Spam is a common side effect of any public widget and does not indicate a security issue with your account.

​
How to handle spam conversations

This is the recommended and most effective workflow when you detect spam in your inbox. Open the spam conversation in your Inbox and click **Assign to me** in the conversation header.
1

Take over the conversation

Guides Handle Spam 1
What “Assign to me” means:
  • You take manual control of the conversation
  • The AI agent stops responding automatically
  • No further AI replies will be sent unless you release the conversation
Use this step as soon as you recognize spam.
2

Close the conversations

When you assign the conversation to yourself, wait until no more spam messages are being sent.
Guides Handle Spammers 1
Once the messages stop, you can remove the conversation from your inbox by clicking on “Close.”What closing does:
  • Removes the conversation from your active inbox
What closing does not do:
  • It does not delete conversation history
  • It does not notify the end user
  • It does not block the user permanently

​
Why this approach works

  • The spammer receives no responses
  • There is no feedback loop for automated scripts
  • Most low-effort spam attempts stop shortly after
The goal is to make your widget an unproductive target.
If the spammer is anonymous, they can:
  • Clear cookies
  • Reload the widget
  • Start a new conversation as a new anonymous user
This is why spam may reappear even after archiving a conversation. There is currently no reliable way to block anonymous users permanently at the widget level.