Claude access reportedly cut by Anthropic for OpenAI

OpenAI has found itself in a challenging situation, as recent reports indicate that Anthropic has terminated OpenAI’s access to its APIs. This action is based on allegations from Anthropic regarding a breach of their terms of service, which has raised significant concerns within the tech community.

As reported by Wired, multiple credible sources state that OpenAI has lost access to Anthropic’s APIs. It has been alleged that OpenAI utilized Anthropic’s Claude Code for assistance in the development and testing phases of their upcoming GPT-5, which is anticipated to be released in August. The implications of this situation are significant for both companies and the broader AI landscape.

Sources suggest that OpenAI was not using the standard chat interface but instead tapping into Claude’s internal tools. This allowed them to use the API for conducting tests on GPT-5, comparing its capabilities in areas such as coding and creative writing against those of Claude. Furthermore, OpenAI reportedly examined safety prompts concerning sensitive topics like CSAM, self-harm, and defamation. Such comparative analysis could provide OpenAI with crucial data to refine GPT-5, enhancing its competitiveness against Claude.

This situation has escalated because it allegedly breaches Anthropic’s commercial terms of service, which explicitly prohibit using their tools to create competing AI products. According to their terms, “Customers may not and must not attempt to access the Services to build a competing product or service, including to train competing AI models or resell the Services except as expressly approved by Anthropic.” This violation has led to serious repercussions for OpenAI.

In response, OpenAI has asserted that their actions align with industry standards, noting that it is common practice for AI companies to test their models against those of competitors. They expressed disappointment over the termination of their API access while acknowledging Anthropic’s right to make such decisions. Interestingly, they highlighted that Anthropic continues to have access to OpenAI’s API.

A spokesperson for OpenAI mentioned that their API access would be reinstated for purposes of “benchmarking and safety evaluations.” This indicates a willingness to collaborate on safety assessments, which are crucial for the development of responsible AI technologies.

This isn’t the first instance in 2023 where Anthropic has restricted API access. In June, they cut off Windsurf’s API access amid rumors of a potential sale to OpenAI. Although that deal ultimately did not materialize, Anthropic’s co-founder, Jared Kaplan, remarked to TechCrunch that it would be unusual for them to sell Claude to OpenAI, highlighting their commitment to maintaining their competitive edge.

Furthermore, Anthropic has also implemented adjustments to its rate limits for Claude, set to take effect in late August. One of the underlying reasons for this change is attributed to a small number of users allegedly violating the company’s policy by sharing and reselling their accounts. This move reflects Anthropic’s efforts to maintain a fair usage policy and safeguard their proprietary technology.

Disclosure: Ziff Davis, the parent company of Mashable, filed a lawsuit against OpenAI in April, claiming that OpenAI infringed upon Ziff Davis copyrights while training and operating its AI systems.