Anthropic revokes API access to OpenAI's Claude service for violating terms of service

AI developer Anthropic has blocked API access to its AI model, 'Claude,' from rival OpenAI, citing a violation of Anthropic's terms of service, which explicitly prohibit the use of Anthropic's AI model to build competing products or services or to train competing AI models.

Scoop: Anthropic revoked OpenAI's API access to its models on Tuesday, multiple sources familiar with the matter tell WIRED. OpenAI was informed that its access was cut off due to violating the terms of service. pic.twitter.com/LyWJAdnY0m

— Kylie Robison (@kyliebytes) August 1, 2025

Anthropic says OpenAI engineers using Claude Code ahead of GPT-5 launch
https://www.bleepingcomputer.com/news/artificial-intelligence/anthropic-says-openai-engineers-using-claude-code-ahead-of-gpt-5-launch/

The issue arose just weeks before OpenAI released its next-generation model, GPT-5. According to the IT news site The Wired , OpenAI engineers were using Anthropic's Claude Code, considered one of the industry's best coding assistants, to test and improve GPT-5's performance, particularly its programming capabilities, which it aims to gain an advantage over. However, rather than using a regular chat screen, OpenAI engineers integrated Claude directly into their internal development tools through an API, systematically comparing and evaluating responses related to coding, creative writing, and even safety.

Anthropic determined that this behavior went beyond mere benchmarking and constituted the development of a competing product, which is prohibited by its terms of service. 'We were not surprised to learn that OpenAI's technical staff were using our coding tools prior to the launch of GPT-5,' said Anthropic spokesman Christopher Nulty, pointing out that this was a direct violation of the company's terms of service.

Sam McAllister, a staff member at Anthropic, also posted on X acknowledging the violation of OpenAI's terms of service.

This take is sad to see but you might not have full context. We cut OpenAl's access for violating our APl terms and for the heavy usage of Claude Code among OAI tech staff.

We're going to continue providing API access for safety evals and benchmarking. That's important to us. pic.twitter.com/VbxbT2Nvbe

— sam mcallister (@sammcallister) August 2, 2025

Meanwhile, OpenAI spokesperson Hannah Wong countered, 'Evaluating competing AI systems is an industry standard to benchmark progress and improve safety,' and pointed out that the company's API is available to Anthropic. 'We are disappointed by Anthropic's decision,' she said.

This is not the first time Anthropic has taken such action. In June 2025, the company blocked access to Claude, a coding tool from Windsurf, which was rumored to be acquired by OpenAI. This series of moves marks a new phase in the AI industry, where competition is heating up and companies are using API access as a strategic weapon to check the moves of their competitors.

This incident could impact the final tweaks to GPT-5, which is rumored to be focusing on improving its coding capabilities. OpenAI may be forced to explore alternatives, such as accessing Claude through a third-party provider, which would increase costs.