AI-Powered Sign-up Fraud Is Scaling Fast

Security experts warn that attackers wielding AI are rapidly shifting their focus to sign-up fraud, gunning for new accounts rather than existing ones. The article, dated 17 September 2025, notes that in 2024, 46% of all registration attempts on the platform were flagged as sign-up attacks, a trend security teams attribute to AI-enabled attack workflows.

It also cites a sustained attack against the retail/ecommerce sector where fraudulent sign-ups outnumber legitimate ones by a factor of 120, turning a promotional incentive into a costly security problem. According to Okta Threat Intelligence, attackers are abusing GenAI tools such as v0, developed by Vercel, to generate phishing sites that impersonate legitimate sign-in pages, lowering the skill barrier for credential harvesting.

The piece also highlights how identity debt and machine-to-machine authentication gaps, including long-lived API keys stored in insecure locations, can enable persistent access if static secrets are exposed. To defend against this, the guidance emphasises strengthening pre-signup rules, using passkeys, and adopting secure OAuth workflows with a central token vault for AI agents.