~/AI SECURITY/inside-the-underground-relay-market-for-discounted-llm-tokens

Inside the Underground Relay Market for Discounted LLM Tokens

An investigation has revealed an underground "relay market" where resellers use open-source proxy tools to sell discounted LLM tokens. These tokens are sourced through API key abuse, exploiting unprotected support bots, free trials, and credit card fraud. This growing fraud vector highlights the security risks for AI developers, who face massive API bills if their public endpoints are exploited to feed this reseller ecosystem. It also pressures LLM providers to implement stricter, real-time spending limits and usage caps on API keys. Resellers leverage legitimate open-source API management tools like `one-api` and `new-api` to load-balance requests across pooled credentials. Buyers use these services to obtain cheap tokens, bypass geographical restrictions, or scrape data for model distillation.

## BACKGROUND

Large Language Model (LLM) providers charge developers based on "tokens" (chunks of text) processed via APIs. Open-source tools like `one-api` were originally designed to help developers aggregate and manage multiple LLM APIs under a single unified interface, but they are now being repurposed by bad actors to pool and resell illicitly obtained API access.

## REFERENCES

## KEYWORDS

#AI Security#LLM APIs#Fraud Detection#Cybersecurity

$ subscribe --daily

Inside the Underground Relay Market for Discounted LLM Tokens | Daily News