A public-facing AI platform just shut down its free trials, and the reason is as concrete as it gets: organised groups are abusing the system for profit. Claire Vo runs ChatPRD, an AI tool that turns product ideas into specs. Since the GPT-4 era, she's built the product in public, but the recent wave of token fraud has forced her to wind down free trials entirely.
The fraud isn't just users trying to stretch a free trial. According to Claire, groups are systematically exploiting promo codes and trial accounts, then tunnelling requests to her agent to run commercial operations. One example: an international Roblox and Fortnite map studio using loops of free trials to generate game maps via her AI backend.
How the abuse works
The attacks are coordinated. Fraudsters spin up new accounts, cycle promo codes, and use the free access to drive real business—either by reselling the tokens or by running their own services through ChatPRD's agent. The scale is enough to swamp a small platform and make offering a free tier financially unsustainable.
Claire notes that some prompts she saw—like SVGs of pelicans riding bicycles—were likely used by attackers to fingerprint which models power the agent. This technique helps scammers identify promising targets or uncover stealth model deployments.
Why shutting down free trials is the only option
The end result: ChatPRD is discontinuing free trials. For an indie AI product, the cost of defending against this kind of fraud—both in terms of engineering time and lost compute—just isn't worth it. The abuse isn't just a nuisance; it directly undermines the business model.
What this means for your own AI service
If you're running or planning to launch any public AI service with a free tier, this is a warning shot. Expect token fraud, plan your defences up front, and be ready to pull the plug on free trials if the economics stop making sense. The attackers are already organised, and they move fast.




