On the surface, companies across nearly every industry seem to be gobbling up AI contracts. Yet under the hood, employees increasingly complain that AI results in more stress while saving no time whatsoever on productive tasks.

Many employers, in turn, have begun mandating AI use on the job, some even going so far as to fire those who don’t hop on board in order to justify their big-time spending on tech industry contracts. While the forced adoption of AI has major implications for the financial viability of the tech overall, it’s also giving office workers a perverse incentive to increase their AI use for non-productive tasks.

Case in point, the Financial Times reports, Amazon’s office staffers are increasingly using the company’s in-house AI agent MeshClaw to run personal tasks in a bid to get their quotas up.

In an attempt to get more than 80 percent of its developers to use AI every week, Amazon has introduced employee-specific AI usage targets in addition to a broader “token consumption” leaderboard that tracks how much each employee uses AI (in machine learning parlance, tokens refer to basic units of data used by AI models to understand text.)

But according to staffers interviewed by the FT, employees are gaming the system by increasingly using the mandated AI systems to automate personal tasks, a tactic known as “tokenmaxxing.”

“There is just so much pressure to use these tools,” one anonymous employee told the paper. “Some people are just using MeshClaw to maximise their token usage.”

For its part, Amazon told the FT that “thousands of Amazonians to automate repetitive tasks each day,” adding that the retailer is “committed to the safe, secure and responsible development and deployment of generative AI for our customers.”

A quick jaunt through Team Blind, a message board for verified employees of companies like Google and Apple, shows that the practice is widespread — or at least, widely acknowledged.

“I burn tokens to s**t my [project manager],” one Amazon employee wrote in a post from May 8. “Whenever my PM says stupid s**t, I launch 10 sub agents to s**t him. Great use of GPUs.”

Asked by a Microsoft worker what that means, the Amazon staffer replied “just paste the Slack conversation history and tell the agent to analyze the guy using 10 sub agents.”

Though less malicious, there are plenty of other threads from workers across the tech industry asking how others are maximizing token usage — receiving lots of free advice in the replies.

“Get creative OP,” one SnapChat worker encouraged under one thread. “If companies use brain-dead metrics to judge people then you need to learn how to f**k them over right back.”

More on AI at work: AI Is Giving Your Boss Tools to Be More Monstrous Than Ever Before