CFPB ordered Block to pay $175M after Cash App's automated system closed disputes uninvestigated
The CFPB found that Block's Cash App relied on an automated macro-based dispute handling system that closed fraud claims without meaningful human review, denied provisional credits required by federal law, and automatically challenged at least 75% of chargebacks without assessing their validity. The consent order filed on January 16, 2025 requires Block to pay $120 million in consumer refunds and a $55 million civil penalty. The violations spanned from 2016 through 2023 and affected hundreds of thousands of Cash App users.
Cash App's automated 'pushback' macros told consumers their case was closed even though the company had failed to investigate at all.
Key facts
- What
- The CFPB found that Block's Cash App relied on an automated macro-based dispute handling system that closed fraud claims without meaningful human review, denied provisional credits required by federal law, and automatically challenged at least 75% of chargebacks without assessing their validity.
- Incident date
- Jan 16, 2025
- Who
- Block, Inc.
- Failure mode
- Agentic Action Error
- AI surface
- Agentic Workflow
- Severity
- High
What happened
The CFPB found that Block's Cash App operated an automated fraud dispute handling system from at least 2016 through 2023 that systematically failed to investigate unauthorized transactions as required by the Electronic Fund Transfer Act and Regulation E. The system used automated macros to close or stall consumer disputes without human review, while auto-denying refunds based on heuristics like 'known device' status rather than the substance of the complaint. Cash App also directed customers to contact their banks to reverse fraudulent transactions while simultaneously blocking those banks' chargeback attempts, and for years provided no live telephone customer support, routing callers to a recorded message instead. At least 153,866 Cash Card claims exceeded the 10-business-day investigation deadline without provisional credits being issued, and the company automatically challenged at least 75% of incoming chargebacks without assessing their validity.
What broke inside the model
- 01 · TriggerAn agent plans a multi-step task.
- 02 · Model stepIt chooses a wrong or destructive action.
- 03 · Control gapNo confirmation gate guards the write.
- 04 · FailureThe action commits to a system of record.
- 05 · ConsequenceData is changed or destroyed irreversibly.
A wrong action commits, and the step is written before anything can stop it.
Cash App's automated dispute system used template macros ('pushback,' 'inquiry,' and 'need more information') to close or stall fraud claims without human investigation, while heuristic rules auto-denied refunds based on factors like whether the transaction came from a 'known device' rather than evaluating the consumer's actual complaint. The system automatically challenged at least 75% of incoming chargebacks without assessing whether the underlying transaction was unauthorized, maintaining an internal 'win rate' metric that prioritized cost avoidance over regulatory compliance. The company also maintained an explicit policy of denying provisional credits for account takeover and peer-to-peer unauthorized transfer claims, even when investigations exceeded the 10-business-day deadline required under Regulation E.
What it cost
Sources
- PrimaryCFPB Orders Operator of Cash App to Pay $175 Million and Fix Its Failures on Fraudconsumerfinance.gov
- Court FilingConsent Order: In the Matter of Block, Inc. (File No. 2025-CFPB-0001)files.consumerfinance.gov
- PressCash App owner to pay up to $175 million for failing to protect customers from fraud, feds saycbsnews.com
Cite this entry
https://failureindex.ai/failures/cfpb-ordered-block-pay-175m-cashAI Failure Index. "CFPB ordered Block to pay $175M after Cash App's automated system closed disputes uninvestigated" (FI-0090). Realm Labs. https://failureindex.ai/failures/cfpb-ordered-block-pay-175m-cash (indexed Jun 4, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0090. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
- AgentRealm
Realm can sit inline on the agent's action path and require that a destructive or high-consequence action clears a real check before it executes, so 'delete and recreate' or a wrong write is stopped at the moment of intent, not explained in the post-mortem.