Financial Times: OpenAI says it has evidence that DeepSeek uses its model for training

On January 29, OpenAI said it found evidence that Chinese artificial intelligence startup DeepSeek used its proprietary model for training.
The company says it has seen some evidence of "distillation," a technique used by developers to achieve better performance on smaller models by using the output of larger, more powerful models, allowing them to achieve similar results on specific tasks at lower cost.
OpenAI declined to comment further on the details of its evidence. Its terms of service state that users cannot "copy" any of its services or "use the output to develop models that compete with OpenAI".
One person close to OpenAI said "distilling" was a common practice in the industry, stressing that the company offered developers a way to do so using its own platform, but said: "The problem is when you're doing this to create your own model for your own purposes."
US AI and crypto czar David Sacks also notes: "There is a lot of evidence that what DeepSeek is doing here is distilling knowledge from OpenAI models, and I don't think OpenAI is happy about that," although he does not provide evidence. (FT)