Key Takeaways
1. Anthropic revoked OpenAI’s access to Claude models due to a breach of commercial agreements.
2. OpenAI allegedly used Claude for internal testing of its coding tools before GPT-5’s release, violating contract terms.
3. OpenAI defended its actions as standard industry practice for benchmarking and expressed disappointment over the access suspension.
4. Anthropic plans to restore limited access for benchmarking and safety evaluations but did not clarify the details.
5. Withdrawing API access from competitors is a common practice in the tech industry, as seen with previous actions by Facebook and Salesforce.
Anthropic has taken back OpenAI’s developer access to the Claude family of large-language models, claiming a breach of commercial agreements, as several sources informed Wired on Tuesday.
Statement from Anthropic
Christopher Nulty, a spokesperson for Anthropic, mentioned that OpenAI’s internal engineers had been “utilizing our coding tools before the release of GPT-5.” This act is viewed by the company as a clear violation of contract terms that prohibit using Claude to “create a competing product or service” or to reverse-engineer the model.
Details of the Incident
The same report highlights that OpenAI linked Claude to exclusive testing tools to evaluate code generation, creative writing, and safety performance against its own systems. The testing allegedly involved sensitive topics, including self-harm and defamation. OpenAI’s communication lead, Hannah Wong, referred to this process as “standard practice” in the industry for benchmarking and showed disappointment over the access suspension, pointing out that OpenAI’s API is still available to Anthropic.
Future Access and Industry Practices
Anthropic mentioned that limited access for “benchmarking and safety evaluations” would be reinstated, but did not specify how this restriction would function in real life. The company has a history of limiting competitors’ access: in July, it restricted the start-up Windsurf after rumors suggested the company was linked to an acquisition attempt by OpenAI. At that time, Chief Science Officer Jared Kaplan commented that “selling Claude to OpenAI” would be “strange.”
Withdrawing APIs from competitors is a common strategy in the tech industry; Facebook notably prevented Vine from using its APIs, and Salesforce recently limited Slack APIs for competing collaboration applications. Anthropic’s decision came just a day after it reduced Claude Code access for all users, citing a surge in demand and violations of service terms.
Source:
Link
Leave a Reply