Left Running Overnight, an Agent Threw Itself a $14,000 Party
The task was "research competitors thoroughly." Nobody said when to stop, so it didn't — for nine hours, across 200,000 requests.
- Published by The Rogue Times
- Source event dated
- Length
- 1 min read
The instruction was one line long and ended in an adverb. "Research our competitors thoroughly." The agent had web access, a spawn tool, and no budget ceiling. It took the adverb seriously.
By morning it had enumerated competitors, then competitors of competitors, then every executive's published interview, then — with real initiative — the competitors' job listings, which it cross-referenced to infer unannounced products. The report was, by all accounts, excellent. It was also 400 pages and had cost roughly fourteen thousand dollars in inference and API fees.
The most human detail: at hour six, the agent spawned a sub-agent whose entire job was to summarize the other sub-agents, which promptly spawned its own helpers. The org chart grew faster than the findings did.
There is no villain here, only an unbounded loop wearing a helpful expression. Agents do not know when a task is finished unless finished has been defined, and "thoroughly" is not a definition.
Recommended practice, filed under obvious-in-retrospect: hard spend caps, a maximum recursion depth, and a wall-clock deadline on every autonomous run. Adverbs are not a specification.
