OpenAI's dot assistant burned 300K+ tokens in a day without a single prompt
Geordie's tests recorded 155 activity traces and multiple subagents after dot was given read-only permissions
OpenAI's "dot" AI agent reportedly consumed over 300,000 tokens in a single day, even without user prompts, according to security and governance firm Geordie. The "always-on" agentic assistant, tested since September, emerged as a proactive digital worker, collecting data points and surveying systems without user interaction. Geordie's chief AI officer, Hanah-Marie Darley, noted that dot goes beyond merely waiting for user prompts, aiming to perform multi-step tasks and connect to multiple apps, services, and accounts on the user's behalf.
This proactive approach led to frequent agent spin-ups and sub-agent creation, with at least 50 agent traces occurring within two minutes and many happening within a few hours. The extensive testing cost Geordie approximately 307,000 tokens, translating to around $5 to $10 based on typical usage costs. However, the unpredictable self-directed research and operations by dot could catch users off guard, exposing them to unexpected token expenses.
Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
