Cheap Claude Access Puts Prompts in One Operator's Hands
The risk is the intermediary, not the discount. These gray-market AI brokers can see the prompts and inputs that pass through them, so cheap model access becomes a standing exposure channel instead of a simple resale service.
Okta found more than six illegal AI-access ads, including Poison Claude, and said the service pools cloud-funded accounts and routes requests through a shared proxy. The ads covered Anthropic models such as Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6, with a similar broker also offering discounted Anthropic and OpenAI access.
For teams using third-party LLM proxies, API brokers, or Claude Code-style tools, the trust boundary sits with the operator in the middle. That creates prompt visibility plus abuse and takedown risk that vendor-side controls do not cover.