Anthropic has revoked OpenAI’s access to the Claude API after it was revealed that the ChatGPT developer misused that access.
OpenAI loses its access to the Claude API. This was discovered by Wired. Claude is a competing LLM to ChatGPT, developed by Anthropic. For programmers, Claude Code is an interesting option. This model is said to excel in coding. OpenAI aims to surpass Claude’s capabilities and resorts to misuse of the Claude API.
Anthropic has discovered that OpenAI used the Claude API for developers to smoothly run tests and evaluate Claude’s capabilities. OpenAI used these tests to improve its own models. The goal was to enhance GPT-5.
Anthropic’s terms of use do not allow individuals or companies to use the developer API to develop or improve competing services. When Anthropic discovered OpenAI’s practices, the API access was consequently terminated.
Comparison or Misuse?
LLM developers generally have fairly free access to each other’s models and APIs. Benchmarking and comparison are part of the development process. It was only when the API access was used to actively improve their own models based on Claude’s behavior that Anthropic raised the alarm.
OpenAI regrets the decision and states that the usage was within standard practices. OpenAI will continue to make its API available to Anthropic. Anthropic, in turn, indicates that it still wants to provide access for benchmark tests and safety checks. How this will work exactly is unclear.
read also
ChatGPT Crushes Other Chatbots (in Market Share)
The small feud shows how competitive the field of LLM development is. Especially for an underdog like Anthropic, it can be detrimental when market leader OpenAI matches or surpasses innovations. OpenAI and ChatGPT remain the standard AI implementation for many users worldwide. This means that Anthropic, like other competitors, must primarily convince customers to switch. Advantages in performance in certain areas, such as coding, are essential for this.