AI free-tier abuse cost calculator
Fake accounts on an AI product don’t just sit in your database. They burn inference you pay for. Estimate how much.
Assumes fake accounts drain 100% of their free credits. Real users use the share you set.
How the calculation works
Monthly burn = signups × fake share × free credits × cost per credit. We assume fake accounts use all their credits, because draining a free tier is the whole point of farming.
Saved with gating is what you would keep by granting flagged accounts only a fraction of the credits until they verify. That’s the limit decision in the BurnGuard API.
Is this a real problem?
Stripe says AI startups with self-serve signups and direct API access “see 10x more attempted abuse than enterprise AI companies”, and in the first two months of its trial-abuse control Radar blocked more than 550,000 high-risk free trials, “preventing an estimated $4.4 million in downstream losses from compute costs”. Stripe
Okta’s threat intelligence team observed over 105,000 signup attempts from 251 IPs against one AI video service’s free trial in a single month, with the farmed credits resold at 5–15% of the official token price. Okta
Read more: AI credit farming, explained and credit gating.
FAQ
How do I estimate my abuse rate?
Count signups from disposable domains, Gmail aliases of existing users and datacenter IPs over the last month, divided by total signups. If you have never measured it, run BurnGuard in monitor mode for a week and read the would-block rate.
What is the cost of a free account?
Free credits granted multiplied by what one credit costs you in inference. For example, 500 credits at $0.002 each is $1.00 per account if the account uses all of it, which farmers always do.
Why assume farmers use 100% of their credits?
Because that is the point of farming. Real users often use a fraction of a free tier; scripted accounts drain it to zero, then resell or consume the output.
Catch these signups automatically.
One API call at signup. Starts in monitor mode, so nothing breaks.