Archive — US — techcrunch
Anthropic cuts off OpenAI’s access to its Claude models
Anthropic has revoked OpenAI’s access to its Claude AI models, citing violations of service terms. OpenAI allegedly used Claude to compare its performance ahead of the GPT-5 launch. Despite this, Anthropic will still allow OpenAI access for safety evaluations. OpenAI expressed disappointment but respects the decision.
Key points
- Anthropic revoked OpenAI’s access to Claude models due to violation of terms of service.
- OpenAI connected Claude to tools for performance comparison in coding, writing, and safety.
- Anthropic stated that OpenAI staff used its coding tools before GPT-5 launch, violating terms.
- Anthropic continues to allow OpenAI access for benchmarking and safety evaluations.
- OpenAI's response highlighted disappointment but acknowledged Anthropic's decision.
Why it matters
- This incident illustrates the competitive tensions in the AI industry as companies seek to protect their intellectual property.
- The restrictions reflect a broader trend within AI development where companies are guarding their technologies amid an arms race in AI advancements.
- Similar situations have previously occurred with tech companies limiting access to proprietary tools to maintain competitive advantages.
The revocation of OpenAI's access to Claude models underscores the ongoing competition in the AI landscape, with companies increasingly protective of their technologies amidst rapid advancements.