[DigitalToday reporter Jinju Hong (홍진주)] Subtle cracks are being detected in the U.S. artificial intelligence (AI) industry. After OpenAI announced a contract with the U.S. Department of Defense, a movement has spread among some users to cancel ChatGPT accounts and move to Anthropic’s Claude.

Claude’s app has recently overtaken ChatGPT in Apple’s App Store download rankings, increasing its presence. Some in the industry say it is not simply a “spillover benefit” but the result of an acceleration of growth that was already under way. Claude ran an aggressive marketing campaign during the Super Bowl with ads that directly targeted OpenAI, and has since stayed near the top of iOS productivity apps. Internal indicators are also rising. Free users rose by more than 60 percent, daily sign-ups tripled and paid subscriptions more than doubled, it was reported.

On the matter, IT outlet TechRadar on March 2 (local time) introduced five key differences that users must know if they plan to switch from ChatGPT to Claude.

First, the way usage limits work on the free plan differs. If ChatGPT is closer to a daily unit, Claude applies limits in 5-hour blocks. The number of times it can actually be used varies depending on message length, whether files are uploaded and server congestion. Users can send roughly a little over 10 messages over 5 hours, but the number can be lower depending on circumstances. If the limit is not enough, users must upgrade to the Pro plan, which costs $20 a month.

Second, longer conversations consume more usage. Claude broadly reflects the full conversation context when generating answers. As a result, token use increases as conversations get longer and users can reach limits faster. Advice has emerged that for prolonged work it is more efficient to start new conversations frequently.

Third, a “memory” feature and data transfer are available. Claude recently introduced a user memory feature even on the free plan. It remembers personalised information such as a user’s job and preferences to provide tailored answers. On paid plans in particular, it also supports a feature that brings over context and settings used on other AI services, a move seen as lowering the burden for existing users to switch.

Fourth, searching past conversations is a strength. Claude can pull up relevant conversations if asked to find earlier chat content. It is seen as useful for document work or research because users do not have to check saved lists one by one.

Fifth, document handling is a strength while image generation is a weakness. The biggest difference cited in real-world use is document processing. Claude is seen as more stable for tasks such as converting image files that include tables into spreadsheet formats or organising complex documents. Some also point out that the completeness of voice mode and image generation features are somewhat lacking compared with ChatGPT.

There are also differences in how they respond. ChatGPT tends to try to produce an answer as much as possible to a given question even if users do not provide enough information, while Claude often asks for additional information or asks again about the purpose of use. This is an approach aimed at securing sufficient context before providing answers, and generally can yield better results.

This shift is drawing attention because it goes beyond a simple comparison of features and reflects user reactions to the values and actions of AI companies. Some also interpret differing views on cooperation with the defense and security sectors as influencing user choices.

Experts, however, say, "We need to watch further to see whether it is a move driven by a short-term issue or a long-term shift in the landscape." In an AI competition where functions, price, ecosystems and ethical stances are all considered, it is still difficult to be definitive about which side will gain the upper hand.