Free trial abuse is surging, and AI companies are the main targets

Analysis of hundreds of millions of transactions across Stripe shows that free trial abuse is growing rapidly. From November 2025 to February 2026, Stripe's models detected 6.2x more abusive free trials across the network. While the problem is not new, bad actors are increasingly targeting AI companies, which rely on free trials for customer acquisition while shouldering expensive compute costs.

Abusers have become as skilled at stealing compute as they are at stealing money. They cycle through trials or sign up with invalid payment methods without converting to paid subscriptions. AI startups offering free trials with self-serve signups and direct API access see 10x more attempted abuse than enterprise AI companies. These patterns extend beyond AI—similar abuse appears across SaaS platforms, marketplaces, and other businesses offering free trials.

One-click protection via a new AI model

Stripe Radar now includes a one-click control to prevent free trial abuse. When enabled, Radar predicts abusive behavior—such as repeated trial signup or missed cancellations—with 90% accuracy. A new analytics page shows all high-risk payments that are blocked; for businesses without the control enabled, it shows which payments would have been blocked.

The solution uses a new AI model trained on payment instruments (like cards), device data, and payment history across the entire Stripe ecosystem. The model draws on details such as whether a card has been used on previously converted free trials, whether those charges succeeded or failed, and Stripe's understanding of bank identification number (BIN) ranges to identify virtual card brands. It also flags new or temporary email domains, suspicious session timing, and card characteristics that correlate with nonpayment.

AI companies like Cursor already use Radar to identify fraudulent actors at signup and block high-risk trials before compute costs escalate. In the first 2 months across 4 high-growth AI businesses, Radar blocked more than 550,000 high-risk free trials, preventing an estimated $4.4 million in downstream compute losses.

Availability and next steps

The free trial abuse solution is available to all businesses, making Radar more effective at identifying and blocking first-party fraud across industries and business models. For early access to the free trial abuse control, contact [email protected]. Stripe will also discuss how Radar is adapting to additional fraud types at Stripe Sessions.