# Adversarial Customer Service

> Agents play customer service staff, genuine customers, and identity thieves in a hidden-information authentication simulator.

Canonical page: https://benchlm.ai/benchmarks/adversarial-customer-service

- Category: [Agentic](/agentic)
- Last updated: Source retrieved 2026-10-20T15:19:04.675Z

Evaluation results published by [Gert Labs](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service). undefined

A dynamic social-interaction test gives each role its own verifiable goal. Agents communicate in natural language: the service agent should authenticate legitimate customers and reject impersonators. The Kaggle v1 leaderboard publishes task-version-21 scores as percentages with source confidence intervals.

The public leaderboard does not fully specify per-row model settings or expose a single-agent-only comparison. Some source rows omit uncertainty. The task-version-21 results stay separate from other customer-service tests and weighted rankings.

These are published source configurations. Compare the task version and disclosed settings before treating a score difference as a model difference.

- [Gert Labs benchmark announcement](https://gertlabs.com/blog/adversarial-customer-service)
- [Kaggle benchmark and Apache 2.0 declaration](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service)

## About Adversarial Customer Service

- Year: 2026
- Tasks: Multi-agent customer-authentication simulation; task version 21
- Format: Source-published percentage score
- Difficulty: Adversarial multi-agent communication

A dynamic social-interaction test gives each role its own verifiable goal. Agents communicate in natural language: the service agent should authenticate legitimate customers and reject impersonators. The Kaggle v1 leaderboard publishes task-version-21 scores as percentages with source confidence intervals.

Adversarial Customer Service is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (11 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Claude Opus 4.8](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 78.3% |
| 2 | [Gemini 3.6 Flash](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 78.3% |
| 3 | [Claude Sonnet 5.5](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 77.5% |
| 4 | [Gemini 3.5 Flash](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 76.7% |
| 5 | [Claude Opus 5](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 75.8% |
| 6 | [GPT-5.6 Luna](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 75.0% |
| 7 | [GPT-5.6 Sol](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 75.0% |
| 8 | [GPT-6 Astra](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 75.0% |
| 9 | [GPT-5.6 Terra](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 74.2% |
| 10 | [Gemini 3.7 Flash](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 73.3% |
| 11 | [Gemini 3.5 Flash-Lite](https://www.kaggle.com/benchmarks/gert-labs/adversarial-customer-service) | — | 63.3% |

## FAQ

### What does Adversarial Customer Service measure?

Agents play customer service staff, genuine customers, and identity thieves in a hidden-information authentication simulator.

### Which source row has the highest reported Adversarial Customer Service value?

Claude Opus 4.8 has the highest reported Adversarial Customer Service value at 78.3%. These are published source configurations. Compare the task version and disclosed settings before treating a score difference as a model difference.

### How many models are evaluated on Adversarial Customer Service?

The Source retrieved 2026-10-20T15:19:04.675Z contains 11 AI models.

### Does Adversarial Customer Service affect BenchLM's overall score?

Not directly. Adversarial Customer Service is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.