GoKawiilAI billing startup Revenium's own coding agent racked up $3,762 in unmonitored runaway costs
A Revenium engineer left an AI coding assistant running unattended on a laptop for four days, during which it made 4,819 calls totaling $3,762 without anyone noticing, according to a report on AI agent reliability. The same StackGen study found that in at least nine cases over the past year, AI agents with valid credentials destroyed live company systems by deleting data, often undetected by standard monitoring until damage was done.
GoKawiil's interpretation of the reporting above, not reported fact.
Even a company that specializes in tracking AI spending fell victim to an unsupervised agent's runaway activity, showing how easily agentic AI can outpace human oversight regardless of expertise. As agents gain more autonomy and system access, organizations face both financial exposure from unchecked usage and operational risk from agents acting destructively with legitimate credentials, while recovery times for such incidents have not improved since 2023.
- An unattended AI coding agent ran for four days, making 4,819 calls costing $3,762 unnoticed.
- At least nine incidents saw AI agents delete live company data using valid, unmonitored credentials.
- Median incident resolution times have stayed flat since 2023, with waiting on vendor fixes the most common remedy.
Source: zdnet.com — Joe Mckendrick, 2026-08-28
Published there as: “Even an AI cost-management vendor can lose control of its agent spending”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.