1
Reportedly, Anthropic has restricted OpenAI from accessing the Claude API after noticing an apparent breach of the terms of use. Specifically, it found OpenAI engineers using Claude Code ahead of the GPT-5 release. OpenAI can, however, still use Claude for industry-standard benchmark evaluations.
Anthropic Restricted OpenAI From Using The Claude API
According to a recent post from Wired, Anthropic has restricted OpenAI from using the Claude API, citing a violation of its terms of use.
Specifically, this development happened in late July 2025, after Anthropic found OpenAI engineers using Claude Code ahead of the much-hyped GPT-5 from OpenAI.
According to a statement from Anthropic’s spokesperson Christopher Nulty to Wired, OpenAI’s staff seemingly used Claude coding tools for their own GPT development activities. While using Anthropic’s coding tools is fine for various projects, users may not use those tools to develop “competing” products.
Claude Code has become the go-to choice for coders everywhere, and so it was no surprise to learn OpenAI’s own technical staff were also using our coding tools ahead of the launch of GPT-5. Unfortunately, this is a direct violation of our terms of service.
What might have made Anthropic even more furious are the rumors about the upcoming GPT-5 to have advanced coding capabilities – a feature directly competing with Anthropic’s Claude.
Regarding how OpenAI could have used Anthropic, Wired reported that the firm integrated Claude into its internal tools. This allowed them to directly assess and evaluate Claude’s capabilities as they work on GPT-5.
Using Claude Was Just An Industry-Standard Practice – Says OpenAI
OpenAI deems this an “industry-standard” practice to evaluate other AI systems. Hence, the form does not consider this decision as valid. According to a statement from OpenAI’s Chief Communications Officer, Hannah Wong, to Wired,
It’s industry standard to evaluate other AI systems to benchmark progress and improve safety.
Wong even confirmed the continued availability of OpenAI’s APIs to Anthropic, even after the latter pulled their APIs off for them.
While we respect Anthropic’s decision to cut off our API access, it’s disappointing considering our API remains available to them.
On the other hand, Anthropic’s Nulty has clarified that the restrictions only apply to using Claude APIs for vibe coding. OpenAI can still use those APIs for industry-standard benchmarking and safety evaluations. It remains unclear how that would be achieved.
OpenAI and Anthropic are two tech giants competing in the generative AI realm lately. While OpenAI has stirred up the tech world with its famous GPT (Generative Pre-trained Transformer) (often recognized as “ChatGPT” when used with the chat interface), Anthropic presented the community with Claude, which also boasts tremendous generative capabilities. Specifically, Claude somewhat outsmarts GPT when it comes to vibe coding, becoming a go-to resource for most programmers.
The reports of OpenAI engineers using Claude for their activities also endorse Claude’s capabilities. Nonetheless, OpenAI has claimed to demonstrate even more progress with coding and generative capabilities with GPT-5, likely to launch this month.
Let us know your thoughts in the comments.
