OpenAI engineers caught using Claude to train GPT-5

Anthropic has restricted OpenAI employees from accessing its AI platform, Claude Code, after discovering it was being used for software development.

The move was reported by WIRED, citing Anthropic spokesperson Christopher Naulty. According to Naulty, using Claude Code to help build a competing product violates the platform’s commercial terms — which explicitly prohibit using the tool for developing rival systems, including model training or reverse engineering.

"Claude Code has become the go-to choice for coders everywhere, and so it was no suprise to learn OpenAI's own technical staff were also using our coding tools ahead of the launch of GPT-5," Naulty added.

OpenAI disagrees with the restriction. Company spokesperson Hannah Wong said testing third-party models is a common industry practice — necessary for measuring progress and improving safety. She also expressed disappointment at Anthropic’s decision, especially given that OpenAI continues to allow competitors full access to its own API.

In response, Anthropic clarified that it won’t completely block OpenAI, but will restrict access to the API for benchmarking and model comparison only — not for programming or internal use.