Anthropic has blocked OpenAI from accessing its Claude models via API, citing a violation of contract as GPT-5’s release nears. According to WIRED, Anthropic alleges OpenAI’s team improperly used Claude Code for internal testing. The commercial terms explicitly prohibit reverse engineering or developing competing products. Anthropic spokesperson Christopher Nulty noted the usage was expected. Reports reveal that OpenAI integrated Claude into its tools for comparisons, testing various applications, including sensitive topics. Anthropic stated it would maintain API access for benchmarking and safety but did not confirm OpenAI’s eligibility going forward. OpenAI’s Hannah Wong called the action “disappointing,” emphasizing the company’s respect for Anthropic’s choice. This isn’t the first instance of Anthropic restricting access post-OpenAI moves; it previously cut Claude access from Windsurf during acquisition talks. With competition in code generation intensifying, Anthropic’s strategies aim to safeguard Claude data from rival advancements, especially as GPT-5 aims to excel in programming tasks.
Source link
Share
Read more