The United States Finance Technology Corporation Slash recently disclosed on social platforms that an employee had accumulated a consumption of about $80,000 when developing a small game using the AI coding tool. The relevant cut-off figure shows that the expenditure was approximately $81,267. The event was followed by a discussion focused on how enterprises can control the cost of the AI tool.

Overconsumption when employees develop games

Slash wrote on platform X that the company had previously encouraged teams to make more use of the AI code, but that an employee accidentally consumed a large amount of money in the production of a game called “Brainrot shot”. The company has also expressed its desire to test the product with the outside world in order to account for its marketing expenses.

In terms of public information, this is a relatively simple-structured shooting game with obvious network engraving elements in the scene and role design. The staff member in question, who issued the N tokens, Las Brilliante, also forwarded the discussion, stating that the situation was “a real accident” and that he had underestimated the intensity of use.

AI, the budget is being revisited.

The incident was rapidly amplified and once again addressed the real problems faced by businesses after the AI tool became available: When the threshold is used down, the test costs may rise rapidly and may not translate into clear outputs.

Business Insider reports that a growing number of companies are revisiting the AI budget. Some enterprises found that the inputs had not led to a corresponding increase in efficiency, and therefore started to reduce expenditures or impose more severe restrictions on the use of staff.

Many companies have a ceiling on their use

It was reported that companies such as Uber, Coinbase and Wal-Mart had capped employee AI expenditure. Among them, Wal-Mart made it clear that this was partly to reduce the need for repetitive “vibe coding”.

Following the rapid entry of the AI tool into the day-to-day office process, similar events have also focused the enterprise's attention on two issues: the authorized boundary and cost recovery. For AI coding applications that are still in the pilot phase, whether the enterprise continues to liberalize access may depend more on actual output and budget constraints.