
Just as OpenAI prepares for the much-anticipated release of GPT-5, Anthropic has taken a significant step by revoking the company’s access to its Claude AI models. The decision was confirmed in a Wired report, citing alleged violations of Anthropic’s terms of service. OpenAI, led by Sam Altman, had reportedly been using Claude via API rather than the standard chat interface to test its capabilities in coding, creative writing, and safety-related tasks.
This access allowed OpenAI to observe how Claude responded to sensitive prompts and benchmark its own models accordingly. Although benchmarking and safety evaluations are typically accepted practices, Anthropic insists the usage went beyond permitted boundaries.
Christopher Nulty, Anthropic’s spokesperson, stated, “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.” He continued, “Unfortunately, this is a direct violation of our terms of service.”
Anthropic Clarifies Scope of Permitted Use
While OpenAI still retains limited access for benchmarking and safety testing, Anthropic emphasized that using Claude to build or train a competing product is strictly forbidden. This clause is clearly outlined in the company’s commercial terms, which prohibit customers from using its services to develop rival AI models or resell the technology.
The decision highlights ongoing tensions between major AI firms as they race to refine their offerings. Furthermore, it raises questions about how companies will handle shared access in a rapidly evolving and competitive industry.
Earlier this year, Anthropic also revoked access for AI coding startup Windsurf amid rumors of an acquisition by OpenAI. At the time, Jared Kaplan, Chief Science Officer at Anthropic, remarked, “I think it would be odd for us to be selling Claude to OpenAI.”
OpenAI Responds to API Cutoff
OpenAI expressed disappointment over the move, although it acknowledged Anthropic’s right to enforce its policies. Hannah Wong, Chief Communications Officer at OpenAI, told Wired, “It’s industry standard to evaluate other AI systems to benchmark progress and improve safety. While we respect Anthropic’s decision to cut off our API access, it’s disappointing considering our API remains available to them.”
Despite the friction, both companies appear committed to upholding industry-wide safety norms. However, with GPT-5’s release looming, competition among top AI developers is clearly intensifying. Whether more access restrictions or collaborations will emerge remains to be seen.