Stripe Data Analysis: Transaction Fraud Down for AI Startups, Multi-Account Abuse Surges

Shift in Fraud Patterns

According to Stripe data, AI startups have been a primary target for fraudsters due to the high resale value of computing resources. In Q3 2025, the transaction fraud attempt rate was up to 4.3x higher compared to all startups, but it decreased to 2.6x in Q1 2026. This is the result of AI companies and Stripe Radar quickly adapting to new fraud patterns.

Surge in Multi-Account Abuse

As transaction fraud success rates declined, actors shifted to Multi-account abuse, repeatedly claiming free trials or new account benefits. From January to June 2026, the fraud attempt rate for AI subscription companies increased by an average of 40%, with the most affected group of companies seeing a 154% increase, and some companies experiencing increases of over 600%.

Stripe Radar's Response

Stripe Radar utilizes billions of transaction signals from the network to assess risk before payment. Specifically, through Multi-account signal and Account sharing signal, it identifies patterns of multiple account creation or concurrent access by a single actor. In the case of ElevenLabs, this feature blocked 2,000 free tier abuses per day over two months.