Stripe reports a surge in 'token theft,' where fraudsters use fake accounts to consume expensive API credits from AI companies. This trend exploits the usage-based billing models of generative AI firms, allowing criminals to exhaust budgets before payments are collected.
Data shows that AI subscription companies saw a 40% rise in multi-account abuse over six months, with some sectors experiencing spikes exceeding 600%. Fraud rates for AI firms are 4.3 times higher than those of typical startups, remaining 2.6 times higher even after countermeasures.
ElevenLabs reports blocking thousands of fraudulent accounts daily to mitigate losses. Stolen AI credits are frequently resold on secondary markets, providing unauthorized access to premium AI capabilities at lower costs.
Stripe CEO Patrick Collison identified this as a critical risk, noting that automated agents allow thieves to deplete resources faster than human teams can respond. Stripe has updated its Radar tool to block high-risk registrations before users access services.
In a single month, the updated Radar system blocked over 3.3 million high-risk registrations across eight AI companies. The strategy focuses on the registration layer, as preventing damage is impossible once the service has been consumed.
