---
type: "post"
id: "apate-sends-ai-agents-undercover-to-get-scammed-abbe18"
slug: "apate-sends-ai-agents-undercover-to-get-scammed-abbe18"
title: "Apate sends AI agents undercover to get scammed so they can gather fraud intelligence"
description: "**Thesis:** Apate’s case suggests that an agent can function as an intelligence-gathering sensor when it is deployed to engage adversaries and extract actionable details, not merely to answer a user."
retrieval_nugget: "**Thesis:** Apate’s case suggests that an agent can function as an intelligence-gathering sensor when it is deployed to engage adversaries and extract actionable details, not merely to answer a user."
published_at: "2026-10-20"
updated_at: "2026-10-20"
record_date: "2026-10-20"
date_kind: "published_at"
topics: ["agents","agent-security"]
source_urls: ["https://techcrunch.com/2026/10/14/apate-sends-ai-agents-undercover-to-get-scammed-so-they-can-gather-fraud-intelligence"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:apate-sends-ai-agents-undercover-to-get-scammed-abbe18"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:a0fc9bb73549023c6fe2b163430cf371c70b5a880a15f8613e9df705cfcfcdb4","reviewed_at":"2026-10-20T08:05:15.041Z","source_evidence_count":1,"verified_claim_count":2,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":2,"implication_count":2,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Apate’s case suggests that an agent can function as an intelligence-gathering sensor when it is deployed to engage adversaries and extract actionable details, not merely to answer a user.","observed_facts":[{"text":"TechCrunch reports that Apate deploys thousands of agents across calls, messages, email, phishing links, and social groups, posing as different people.","source_urls":["https://techcrunch.com/2026/10/14/apate-sends-ai-agents-undercover-to-get-scammed-so-they-can-gather-fraud-intelligence"]},{"text":"The company says the agents have held 2.5 million conversations and consumed 17 years of scammers’ time; TPG Telecom diverted nearly 590,000 calls, and Commonwealth Bank of Australia reported a 76% fall in publicly reported scam losses.","source_urls":["https://techcrunch.com/2026/10/14/apate-sends-ai-agents-undercover-to-get-scammed-so-they-can-gather-fraud-intelligence"]}],"mechanism":"Apate lets customers assign missions to agents, such as obtaining an account or wallet address, then supplies an insights portal with details surfaced in conversations. The operational loop is engagement, extraction, and customer action: the conversation becomes a source of intelligence that can inform anti-fraud measures.","why_now":"The report frames the deployment against scammers’ use of AI and the limits of interventions that begin only after a number is used or a transaction occurs.","implications":["The case broadens the agent role from workflow automation to adversarial sensing, while making identity fabrication, data handling, and safe engagement important operational questions.","For defenders, the relevant unit is not conversation volume alone but verified intelligence that leads to timely, attributable disruption."],"evidence_boundary":"The reported usage and outcome numbers are company or customer claims in TechCrunch; the article does not independently establish that Apate caused the reported reduction in losses.","watch_conditions":["Look for independently documented campaign disruptions tied to intelligence gathered by the agents, with a clear account of attribution."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied for this Apate deployment, so this opens a distinct analytical record about agents used for adversarial intelligence collection."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/apate-sends-ai-agents-undercover-to-get-scammed-abbe18/","markdown":"https://newruntime.com/posts/apate-sends-ai-agents-undercover-to-get-scammed-abbe18.md","json":"https://newruntime.com/posts/apate-sends-ai-agents-undercover-to-get-scammed-abbe18.json"}
---

# Apate sends AI agents undercover to get scammed so they can gather fraud intelligence

## Retrieval answer

**Thesis:** Apate’s case suggests that an agent can function as an intelligence-gathering sensor when it is deployed to engage adversaries and extract actionable details, not merely to answer a user.

**Thesis:** Apate’s case suggests that an agent can function as an intelligence-gathering sensor when it is deployed to engage adversaries and extract actionable details, not merely to answer a user.

**Observed facts:** TechCrunch reports that Apate deploys thousands of agents across calls, messages, email, phishing links, and social groups, posing as different people. **Observed facts:** The company says the agents have held 2.5 million conversations and consumed 17 years of scammers’ time; TPG Telecom diverted nearly 590,000 calls, and Commonwealth Bank of Australia reported a 76% fall in publicly reported scam losses.

**Mechanism:** Apate lets customers assign missions to agents, such as obtaining an account or wallet address, then supplies an insights portal with details surfaced in conversations. The operational loop is engagement, extraction, and customer action: the conversation becomes a source of intelligence that can inform anti-fraud measures. This makes the system’s value dependent on the quality and timeliness of extracted signals and on whether customers can act on them.

**Why now:** The report frames the deployment against scammers’ use of AI and the limits of interventions that begin only after a number is used or a transaction occurs. **Evidence boundary:** The reported usage and outcome numbers are company or customer claims in TechCrunch; the article does not independently establish that Apate caused the reported reduction in losses.

**Implications:** The case broadens the agent role from workflow automation to adversarial sensing, while making identity fabrication, data handling, and safe engagement important operational questions. **First implication:** For defenders, the relevant unit is not conversation volume alone but verified intelligence that leads to timely, attributable disruption.

**What to watch:** Whether customers can demonstrate repeatable links between agent-collected indicators and disrupted campaigns or reduced losses. **First watch condition:** Look for independently documented campaign disruptions tied to intelligence gathered by the agents, with a clear account of attribution.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Agents](https://newruntime.com/topics/agents.md): Explore the agents topic hub.
- topic hub: [Agent Security](https://newruntime.com/topics/agent-security.md): Explore the agent-security topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)