Amazon employees reportedly pretending to be more AI-dependent than they really are

There was a time when employees exaggerated how busy they were by keeping spreadsheets open during lunch breaks or typing furiously when a manager walked past. Inside Amazon, some workers are reportedly performing a more modern version of the same ritual: making themselves look deeply dependent on artificial intelligence.

According to a report from the Financial Times, employees at Amazon have begun using the company’s internal AI systems to automate additional or unnecessary tasks, partly to create the impression that they are heavy AI users.

The behaviour points to an increasingly unusual dynamic taking shape across the technology industry. Workers are not simply being encouraged to use AI tools anymore. In some environments, they increasingly feel judged on how visibly and enthusiastically they use them.

At the centre of Amazon’s internal AI push is a tool called MeshClaw, software that allows employees to create AI agents capable of interacting with workplace applications and carrying out tasks automatically.

More from Tech

Amazon employees are reportedly using AI to look more AI-native

MeshClaw functions less like a simple chatbot and more like an automated digital assistant.

The tool can reportedly handle tasks such as sorting emails, monitoring deployments, interacting with Slack and initiating certain coding workflows. Internal descriptions portray it almost like an always-on co-worker operating quietly in the background.

One internal memo reportedly described the system as something that “dreams overnight” while consolidating information and preparing tasks before employees begin work the next day.

But several employees told the Financial Times that some workers are now using the software beyond genuine productivity needs. Instead, they are generating additional AI activity simply to increase their AI token usage, the measurement system tracking how much employees engage with AI models internally.

The trend has reportedly become significant enough to earn a nickname inside the tech industry: “tokenmaxxing”.

Internally, Amazon has introduced goals encouraging widespread AI adoption.

More than 80 per cent of developers are reportedly expected to use AI tools weekly, while internal leader boards have tracked AI token usage across teams.

Although Amazon has told employees the figures will not directly affect performance evaluations, several staff members reportedly believe managers are still monitoring the numbers closely.

That perception appears to be shaping behaviour across teams.

Why Silicon Valley wants employees to embrace AI so aggressively

The pressure surrounding AI adoption reflects broader anxieties inside the technology industry.

Companies such as Amazon, Meta and others are investing billions into generative AI infrastructure and tools. Amazon alone is expected to spend roughly $200 billion this year, with much of that focused on AI systems and data centres.

Internally, these companies want employees to integrate AI deeply into daily workflows, partly to increase efficiency and partly to demonstrate that expensive AI investments are producing measurable cultural change. Meta has also introduced a software to monitor how employees perform their task, so as to feed the same to AI and hence the tasks can be automated.

Amazon says MeshClaw is helping thousands of employees automate repetitive work and experiment responsibly with AI technologies.

Still, not everyone inside the company appears convinced.

Several employees reportedly raised concerns about the security implications of AI agents being allowed to act independently on a user’s behalf. Tools capable of deploying code, interacting with workplace systems or managing communications autonomously may also create new risks if mistakes occur or unintended actions are triggered.

More broadly, some workers appear uneasy about what the rise of AI metrics means for workplace culture itself.

In earlier eras of Silicon Valley, employees competed over productivity, innovation or long working hours. Increasingly, they may now also be competing over how convincingly they can appear inseparable from AI.