Mule Harvest: scam-baiting agents that collect mule accounts for banks
AI agents engage live scammers to collect the receiving accounts and wallets they use. Bank and credit union fraud teams pay.
Competitive · Scams take 22% of fraud dollars
Written for: Small startup team
Next step
Call the fraud or BSA lead at a $2B–$10B credit union with rising authorized-scam losses. Ask whether a weekly list of live mule accounts held at their own institution would change how they hold and review outgoing payments.
Why now
PYMNTS reports that scams now account for 22.4% of fraud dollars, and many of these payments come with a valid login or the customer's own approval. That means login and device checks miss them. What banks lack is information about where the money is going: the mule account numbers and wallet addresses scammers are using this week. Apate has proven the method in Australia, but US credit unions and mid-size banks have no feed of their own.
What became possible
Apate.ai shows that agents can hold convincing conversations with scammers long enough to get payment instructions out of them. Cheap chat and voice models make it affordable to run thousands of these conversations at once.
Two ways in
- Mule Harvest feed
Agents respond to scam texts, romance-scam profiles and fake-investment ads, and log every bank account, routing number and wallet address they are told to pay. Each bank gets an alert when one of its own accounts appears, so it can freeze a mule, and a list of external payees for holds on outgoing payments.
- Customer
- Fraud and BSA teams at US credit unions and regional banks
- Model
- Annual subscription per institution, tiered by asset size
- Wallet watch for issuers and exchanges
The same collection pipeline limited to crypto addresses, delivered as a watchlist to exchanges and stablecoin issuers that can freeze funds.
- Customer
- Compliance teams at exchanges and stablecoin issuers
- Model
- Subscription per API call or per seat
Competition
The core idea is already sold. Netcraft, CYBERA, Cube AI and Apate.ai all run AI agents against live scammers and sell the collected mule accounts and wallets to banks, exchanges and telecom companies, and Cube is already working with Navy Federal in the US.
Full competition notes (4 competitors)
The core idea is already sold. Netcraft, CYBERA, Cube AI and Apate.ai all run AI agents against live scammers and sell the collected mule accounts and wallets to banks, exchanges and telecom companies, and Cube is already working with Navy Federal in the US. A new entrant cannot win on the basic idea or on having the data. Its best chance is the long tail of small US credit unions and community banks, which the incumbents seem to underserve. That would mean a cheap self-serve alert product that matches each institution's own routing numbers to flag its mules and gives an outside payee list for outgoing holds. It could be delivered through the core banking and fraud-monitoring systems those teams already use, or through a credit union service organization or consortium, and could focus on US channels like Zelle, P2P apps and check fraud. The crypto-only watchlist faces Netcraft and Apate, which already collect wallets, and probably also blockchain analytics firms like Chainalysis and TRM, which this check did not research. That makes it the weaker of the two ideas unless the new entrant has unusually fresh, high-volume collection.
- Netcraft Conversational Scam Intelligence
Uses AI agents in private messages with scammers to collect mule bank accounts and crypto wallets. Reports thousands of mule accounts across 73 countries and more than 600 institutions, plus wallets that received over $45M. Customers use the data to flag or block payments, and Netcraft pitches it to exchanges too. This is the closest match to both product ideas.
- CYBERA
Agentic personas engage live scammers. Claims more than 16,000 confirmed mule accounts in 72 countries in the second half of 2025. Data goes into bank and insurer payment screening, it partners with Recorded Future, and it raised $2.5M. Its authors published an arXiv evaluation of the method (https://arxiv.org/abs/2509.08493v1).
- Cube AI (Navy Federal Credit Union)
Bots converse with scammers and pass the receiving accounts and payment handles to financial institutions. Navy Federal, the largest US credit union, has used it since January to interdict scam payments and warn members.
- Apate.ai
Australian startup running thousands of scam-baiting bot profiles for Commonwealth Bank. Outputs mule accounts, crypto wallets, phone numbers and scam scripts. Has a global telecom deal with Mobileum and is expanding to London. No US bank customer found.
Checked 2026-10-16
Risks
- Apate leads and may enter the US first
- Legal and ethical questions around deceiving criminals, recording calls and sharing data between institutions, so the 314(b) information-sharing route needs a lawyer
- Scammers learn to spot bots, and mule accounts are swapped quickly, so the data goes stale fast
- Bank procurement cycles are longer than 3 months
Best on tech readiness and market pull; open space is the weak point (5).
tech 7 · pull 7 · space 5 · fit 7 · build 7
Exposed
Companies that could benefit or be threatened. News and context, not investment advice.
Momentum of its main topic
Sources
Status history
- 2026-10-16 Flagged as buildable