[Digital Today intern reporter Seung-ah Yoo] Some users moved from ChatGPT to Claude after OpenAI cooperated with the U.S. Defense Department, but there is a disappointed reaction to usage limits that are stronger than expected.

TechRadar, an IT media outlet, reported on March 9 that some users have been leaving ChatGPT and moving to Claude after it became known that OpenAI cooperated with the U.S. Defense Department. It also said this was influenced by Anthropic being classified as a "supply chain risk company" after friction with the U.S. Defense Department.

Claude’s user numbers have risen rapidly. In the United States, it surpassed ChatGPT in downloads, posting 1.1 million downloads in February, up 240 percent from the previous month. It also ranked No. 1 among free iPhone apps and was found to have secured more than 1 million new sign-ups a day.

ChatGPT users disappointed by OpenAI’s cooperation with the Defense Department showed a movement to shift en masse to Claude. As a result, it surpassed ChatGPT in downloads in the United States and recorded 1.1 million downloads in February, up 240 percent from the previous month. It also took the top spot among free iPhone apps and secured more than 1 million new sign-ups a day.

Online interest also surged in posts summarising "what to know before switching from ChatGPT to Claude". Anthropic also moved to attract users by offering a memory feature to free users and supporting the use of data saved in other AI chatbots by importing it.

But the experience of newly arriving users differed from expectations. Complaints are continuing as Claude’s free usage limits are applied more strictly than expected. ChatGPT users used to a relatively generous environment, in particular, are expressing disappointment at Claude’s limited usage.

Claude operates with multiple models like ChatGPT. It ranges from lightweight models to high-performance models, with Haiku, Sonnet and Opus among the main ones. The higher the performance, the faster usage is consumed.

As with most artificial intelligence services, the free plan sharply limits functions and usage. The problem is that usage is not sufficient even on the paid plan of $20 a month. Opus, the highest-performance model, can use up a day’s allowance with about 10 to 15 conversations. An assessment has emerged that the model most users can realistically use is Sonnet, a mid-tier option.

To secure more usage, users must subscribe to the $100-a-month "Max" plan. This plan provides far more usage than the API method, but it is a burdensome price for ordinary users.

This structure is also related to Anthropic’s business strategy. From the start, Claude was designed as a service centred on developers and corporate customers rather than the mass market. It is a strategy aimed at securing a small number of high-paying customers. OpenAI, by contrast, has expanded its business with a mass-market platform aimed at securing 1 billion users in the long run.

Some experts say Claude’s usage limits could instead produce positive effects. That is because unlimited access to AI could cause excessive dependence or addiction. Recently, emotional dependence on AI chatbots has also emerged as a social issue.

Of course, Anthropic did not limit usage out of concern for users’ health. It was a decision driven by costs and business strategy. Still, an assessment has emerged that a commercial choice could produce unexpected positive effects. Some experts also predict that many users may ultimately return to ChatGPT.