OpenAI was blocked from API access after it was revealed that the company illegally utilized code from competitor Anthropic's artificial intelligence, Claude, during the development of its next-generation model, GPT-5. There are indications that this case is similar to one involving developers at the Chinese company DeepSeek, who distilled OpenAI data and became embroiled in controversy.
According to industry sources on the 13th, Anthropic confirmed just after discovering that OpenAI developers had linked Claude to its internal system via the developer API, using it directly for coding, creative writing, and safety testing, and subsequently blocked access. It is reported that OpenAI systematically analyzed Claude's responses to sensitive prompts regarding child sexual exploitation, self-harm, and defamation to reflect improvements in the safety of GPT-5.
Anthropic stated that using the API for developing competing products, training models, and reverse engineering or replication is a clear violation of its terms. In contrast, OpenAI expressed regret, stating that measuring performance and enhancing safety using other AI systems is an industry standard. The industry views the direct integration of competitor technology into model development systems as fundamentally different from simple comparisons.
Some suggest that the enhanced coding capabilities of GPT-5 may reflect characteristics of Claude's code. Experts assess that this incident, similar to the recent DeepSeek controversy, has blurred the ethical and legal boundaries surrounding the use of competitor technology, intensifying conflicts.