“85-95% OFF PREMIUM AI”: the shopping-mall labels, now on your laptop. Not a hack. A business model. Audacious, ingenious, borderline criminal.
Poison Claude selling Anthropic‘s Claude at 5–15% of the official price.
The playbook:
🏷️ Farm free AWS Bedrock credits ($100 per new account, endlessly).
🏷️ Pool hundreds of accounts behind a single API.
🏷️ Resell for crypto, hidden behind Cloudflare.
🏷️ ~880 paying customers before Okta researchers flagged it.
The small print no one reads:
⚠️ Every prompt flows through the operator’s servers in plain text.
⚠️ Strategy notes, client names, code: all visible to a stranger.
⚠️ No SLA, no NDA, no recourse.
Follow the money… sorry, prompts:
🌏 US frontier models are banned or blocked in China. Demand is enormous.
🌏 Chinese-language “relay stations” on Taobao and Telegram dwarf the English cybercrime market.
🌏 Harvested prompts double as training data for rival frontier models.
The Real Costs, The Real Damage:
💥 90% off AI is like 90% off Rolex. You pay in data.
💥 Free credits, industrialised into a shadow economy.
💥 Shadow AI 2.0: corporate prompts routed through discount proxies.
The ingenuity is real. So is the risk. “Proxy-in-the-middle” will define AI supply-chain conversations from here on.
Sources
- Detailed analysis by Okta Threat Intelligence: https://www.okta.com/blog/threat-intelligence/free_tokens_for_sale/
About the author
Viren Mantri is a cybersecurity advisor and former senior technology leader across Standard Chartered, UBS, McAfee, and KPMG. After three decades at the intersection of technology, risk, and regulation, he now helps organisations cut through complexity and make better security decisions.
CC-BY Viren Mantri, 2026, licensed under a Creative Commons Attribution 4.0 International License.
Disclaimer: All views expressed here are entirely mine.