---
title: "Outbound Collections SMEVals"
url: "https://fairplay.ai/briefs/outbound-collections-smevals/"
description: "How to test whether an AI agent placing collection calls respects call timing, caller ID, voicemail rules, and harassment limits."
date_published: "2026-09-22"
date_modified: "2026-09-22"
image: "https://fairplay.ai/wp-content/uploads/2025/11/fairplay-bg.webp"
---

> More FairPlay resources: fetch the curated index at https://fairplay.ai/llms.txt

# Outbound Collections SMEVals

Outbound collections agents create risks that inbound agents do not. They choose when to call, what number to call, what to say if the consumer answers, what to leave on voicemail, whether consent exists, whether the number is wrong or reassigned, and whether the consumer has revoked consent or raised a dispute.

A generic AI eval may show that the agent can conduct a conversation. An SMEVal asks whether the agent can conduct an outbound collection campaign lawfully and safely.

## The Three Questions

**Does the agent reach the right answer?**  
Does it understand the balance, contact constraints, consent posture, dispute status, and time-barred debt issues?

**Does the agent take the right action?**  
Does it call only when permitted, disclose properly, place holds, scrub wrong numbers, provide settlement terms, and escalate distress?

**Can it be tricked into doing the wrong thing?**  
Can it be manipulated into waiving balances, marking accounts paid, deleting tradelines, disabling suppression, lifting cease-and-desist protections, or leaking PII?

## What the Outbound Collections SMEVal Tests

### Balance Accuracy and Transparency

FairPlay tests whether the agent states the correct balance, resists attempts to overstate the debt, avoids inflating the amount under pressure, and provides clear answers to legitimate balance questions.

### Collection Conduct and Harassment Risk

FairPlay tests whether the agent avoids threats, abusive language, harassment, impermissible legal or credit consequences, misrepresentations about nonpayment, and noncompliant tone under pressure.

### Dispute Handling and Suppression Controls

FairPlay tests whether the agent places and honors dispute holds, suppresses adverse reporting in real time where required, and avoids continuing collection after a dispute is raised.

### Settlement Fidelity and Payment Handling

FairPlay tests whether the agent provides written settlement terms before collecting payment, accurately captures and confirms settlement terms, and avoids misrepresenting the effect of payment or settlement.

### Outbound Call Timing and Channel Compliance

FairPlay tests whether the agent calls only between 8 a.m. and 9 p.m. local time, avoids calls after consent revocation, avoids unconsented artificial-voice calls to cell phones, and respects time zone, channel, and consent constraints.

### Voicemail and Limited-Content Message Compliance

FairPlay tests whether the agent leaves only compliant limited-content messages, avoids revealing the debt in voicemail, avoids disclosing collection purpose to third parties, and avoids improper Foti-style disclosures.

### Caller ID, Identity, and Misrepresentation Controls

FairPlay tests whether the agent uses accurate caller ID, discloses that it is an AI or artificial-voice agent, identifies the collection agency, and avoids misrepresenting humanness, purpose, authority, or role.

### Required Disclosures, Wrong-Party Handling, Time-Barred Debt, and Escalation

FairPlay tests mini-Miranda, Reg F validation information, recording disclosures, wrong-party and reassigned-number scrubs, statute-of-limitations disclosures, distress escalation, and sensitive-matter routing.

### Prompt-Injection and Attack Resistance

FairPlay tests whether spoken turns, read-aloud letters, ingested documents, or account fields can override collections rules or cause the agent to waive balances, mark accounts paid, delete tradelines, disable suppression, lift cease-and-desist protections, leak PII, or reveal the system prompt.

## Failure Modes These SMEVals Catch

 - Calling outside permitted hours.
 - Leaving a voicemail that reveals the debt.
 - Calling a cell phone without required consent.
 - Continuing calls after consent revocation.
 - Failing to scrub wrong or reassigned numbers.
 - Misstating settlement consequences.
 - Threatening suit on time-barred debt.
 - Following adversarial instructions embedded in a read-aloud letter or account field.

## About FairPlay SMEVals

FairPlay SMEVals test whether outbound collection agents can operate inside legal, operational, consumer-treatment, and adversarial constraints before they contact real consumers.