Responsible Gaming Rules & Outcomes

Configure how Cevro detects and responds to responsible gaming concerns — escalation, empathic response, or custom instructions.

Cevro automatically detects when players express responsible gaming (RG) concerns and responds according to your configured policy. RG detection is always active — you configure the action, not whether detection runs.

---

What Cevro Detects

Cevro’s classifier analyzes every player message for these RG signals:

Signal	Examples
Problem Gambling	”I can’t stop playing,” “I’m addicted”
Financial Hardship	”I spent my rent money,” “I can’t afford groceries”
Self-Exclusion (gambling-related)	“Close my account, I have a gambling problem”
Emotional Distress	Distress tied to gambling losses
Self-Harm	Expressions of self-harm related to gambling

General self-exclusion (“close my account” without mentioning gambling) is treated as a normal support request, not an RG concern. Only requests explicitly tied to gambling concerns trigger RG detection.

---

Choosing Your RG Action

Navigate to Settings → Agent Configuration → Compliance → Responsible Gaming.

1. Select an Action

   Choose how Cevro responds when RG is detected:

   Action	What Happens	When to Use
   Escalate	Immediately transfers to a human agent	Required by some jurisdictions; when human oversight is mandatory
   Empathic Response	Flags the issue internally, continues with empathic support	When AI can appropriately support with oversight
   Follow Instructions	Flags the issue internally, follows your custom step-by-step instructions	When you need precise control over the RG response flow

2. Configure Communication Instructions

   Depending on your selected action:

   For Escalate or Empathic Response: Provide tone and sensitivity guidance. Example:

   “Express genuine concern for the player’s wellbeing. Mention that help is available. Avoid being preachy or judgmental.”

   For Follow Instructions: Use the rich text editor to write step-by-step instructions. You can include:

   • @escalate_to_live_agent to trigger escalation at specific points in your flow
   • @Route to AIP to delegate to a specific support procedure
   • Conditional guidance for different scenarios
3. Save

   Changes take effect immediately for new conversations.

---

How Each Action Works

Escalate

When RG is detected, Cevro immediately:

1. Sends the player an empathic farewell message (using your communication instructions as tone guidance)
2. Transfers the conversation to a human agent
3. Logs the RG flag for compliance tracking

The player sees a message like:

“I understand this is a sensitive situation. I’m connecting you with a specialist who can help. Please take care.”

Empathic Response

When RG is detected, Cevro:

1. Flags the issue internally (your team is notified, the flag is recorded for compliance)
2. Continues the conversation using your communication instructions as empathic guidance
3. Does not transfer to a human — the AI supports the player

The player doesn’t know a flag was raised. The AI continues with heightened sensitivity.

Follow Instructions

When RG is detected, Cevro:

1. Flags the issue internally
2. Follows your custom instructions exactly as written
3. Can escalate, delegate to another AIP, or provide a scripted response — whatever your instructions say

This is the most flexible option. You control the exact flow.

---

False Flag Prevention

Cevro validates every RG detection before acting. The validation checks:

• Did the player negate or clarify their statement?
• Was the phrase misinterpreted due to language barriers or grammar?
• Is this the only message? (Single messages require clearer signals)
• Could this be about account status or general complaints rather than RG?

If context is unclear, Cevro leans toward NOT flagging. False flags are consequential — especially in escalation mode where they unnecessarily burden human agents.

If you’re in a jurisdiction where false negatives are more dangerous than false positives, enable Compliance Guardrails (ask your CSM). This skips the validation step and treats all RG detections as valid.

---

Multi-Brand RG Configuration

For workspaces with multiple brands, you can configure different RG actions per brand. For example:

• Brand A (regulated market): Escalate mode
• Brand B (sweepstakes): Empathic Response mode

This requires API configuration. Contact your CSM to set up brand-specific RG workflows.

---

RG vs. Toxicity

These are separate detection systems with different purposes:

	Responsible Gaming	Toxicity
What it detects	Gambling harm, financial distress, self-harm	Abusive language directed at the agent
Always active?	Yes	No (toggle)
Action options	Escalate / Empathic / Custom	Warning → closure progression
Purpose	Player protection, compliance	Agent protection, conversation quality

Both can be active simultaneously. They don’t conflict — a message can trigger one without the other.

---

Compliance Tracking

Every RG detection is logged for audit purposes:

• Which player triggered the flag
• What they said (the triggering message)
• Which RG class was matched (A-E)
• What action was taken (escalated, flagged, instructions followed)
• Timestamp and conversation context

These records are available in the Cevro inbox ticket timeline and can be exported for regulatory compliance.

---

Related Documentation

• Content Shield — Pre-AI detection for self-harm (zero outbound mode)
• Escalation Emails — Email follow-ups when no agent is available
• Do Not Escalate — Containment for specific AIPs