Anthropic terminated OpenAI's access to its Claude artificial intelligence models after discovering terms of service violations. The competitor had been using Claude's capabilities to benchmark and improve its GPT systems. OpenAI accessed multiple Anthropic tools through developer interfaces to enhance GPT-5 development. Claude Code served as a primary resource for OpenAI's technical teams. Commercial agreements explicitly prohibited competitors from using Claude data to train rival models.
Anthropic maintained limited access for safety testing and performance evaluations despite the broader restrictions. OpenAI expressed disappointment with the decision while noting that Anthropic continues accessing OpenAI's services. The access termination followed OpenAI's acquisition of Windsurf, a coding startup that concerned Anthropic executives. This competitive dynamic mirrors previous industry actions where companies restricted data sharing with rivals. Anthropic officials worried that continued access would allow OpenAI to develop superior coding and writing capabilities.
Anthropic maintained limited access for safety testing and performance evaluations despite the broader restrictions. OpenAI expressed disappointment with the decision while noting that Anthropic continues accessing OpenAI's services. The access termination followed OpenAI's acquisition of Windsurf, a coding startup that concerned Anthropic executives. This competitive dynamic mirrors previous industry actions where companies restricted data sharing with rivals. Anthropic officials worried that continued access would allow OpenAI to develop superior coding and writing capabilities.
