They spent $500 million on a chatbot. Accidentally

Running any business involves unforeseen costs - including repairing unwise mistakes made by employees. Printers with incorrectly replaced ink, unclean coffee machines, or accidental disposal of documents. The long list of possible carelessness and mistakes now includes the possibility of burning a fortune talking to chatbots.

Corporate AI fever can be expensive. Here's an example

According to information published by Axios, an anonymous company was to spend approximately USD 500 million (approximately PLN 1.816 billion) in one month on using Claude, an AI model developed by Anthropic. The huge bill was not the result of a hacker attack or system failure. The source of the problem turned out to be the lack of usage limits for employees.

A consultant dealing with artificial intelligence implementations, quoted by Axios, talked about the matter. According to his account, the company made Claude licenses available to employees without appropriate limits on the number of tokens used. The consequences of such a decision quickly translated into astronomical costs.

The company's name was not disclosed, but the amount itself caught the attention of the technology industry. Half a billion dollars spent in one month is an amount comparable to the budgets of large infrastructure projects or the costs of acquisitions of smaller technology companies.

The story was cited by Axios as an example of the problems that enterprises begin to face after a period of intensive implementation of AI tools. Many companies are currently trying to assess whether high spending on chatbots, AI agents and other solutions using large language models translates into tangible business benefits.

The restrictions were decided by Microsoft, which imposed limits on the use of part of the Claude Code license by employees. Uber's chief operating officer, in turn, allegedly stated that the costs associated with artificial intelligence were becoming increasingly difficult to justify.

Experts also point to the problem of inappropriate use of available tools. One technology executive quoted by Axios admitted that employees even used AI models to check the weather forecast. Each such inquiry generates costs that, if large enough, can significantly burden the company's budget.

Sophia Velastegui, former director of AI at Microsoft, estimated that many employees automate tasks they don't like doing, instead of those that bring the greatest value to the company. This strategy may increase the use of AI tools without a commensurate increase in benefits.

Rising bills also remain a challenge for language model providers themselves. Maintaining the infrastructure needed for the operation of advanced AI systems requires huge financial outlays, which is why some companies increase the prices of services or tighten usage limits - not only for free users, but also for paying ones.