It has been discovered that Amazon employees are inflating their AI usage to improve their rankings and impress their superiors.



An Amazon employee shared a story about unnecessarily continuing to use the company's AI tools to improve their image with management.

Amazon employees are inflating AI usage to top leaderboards and impress managers | TechSpot

https://www.techspot.com/news/112386-amazon-employees-using-internal-ai-tools-inflate-usage.html



Amazon employees are 'tokenmaxxing' due to pressure to use AI tools - Ars Technica

https://arstechnica.com/ai/2026/05/amazon-employees-are-tokenmaxxing-due-to-pressure-to-use-ai-tools/

Amazon has developed MeshClaw, an internal AI platform inspired by OpenClaw , which provides a mechanism for running local AI agents on its own hardware.

According to three Amazon employees, some employees are using MeshClaw to spend as many tokens as possible and perform unnecessary tasks in order to make a good impression on management.

Sources say, 'This move comes after Amazon set a goal for more than 80% of developers to use AI weekly and began tracking token usage in internal rankings.'



Amazon initially shared team-level statistics on MeshClaw usage, but later restricted access to internal staff only. Officially, high token consumption does not affect employee performance evaluations, but unofficially, some managers monitor token consumption and use it to their advantage in performance reviews. Knowing this, some employees are inflating their MeshClaw usage, leading to competition amongst themselves.

One source said, 'There is a lot of pressure to use these tools. Some people are using MeshClaw simply to maximize their token usage.'

Amazon states that 'thousands of employees currently use MeshClaw on a daily basis.'

A source said, 'Some employees are uneasy about using generative AI to automate large parts of their work, particularly due to concerns about hallucination and factual inaccuracies. They are terrified of the default security posture and do not want to let the AI run wild and do whatever it wants.'

in AI, Posted by log1p_kr