Published on: February 1, 2025
OpenAI is accusing Chinese AI startup DeepSeek of using its proprietary models to train its open-source competitor, pointing to a possible intellectual property breach. OpenAI says DeepSeek used a technique called distillation to train its AI model.
In machine learning, distillation is a technique where a smaller model is trained to mimic the performance of a larger, pre-trained model by learning from its outputs. This allows developers to create more efficient AI systems with similar capabilities while using fewer resources. For instance, OpenAI spent over $100 million to develop GPT-4.
Although OpenAI permits access to its API for integration into third-party applications, using its outputs to build rival AI models violates its terms of service. The company claims to have evidence linking DeepSeek to this practice but has not disclosed specific details supporting its allegation.
“The issue is when you [take it out of the platform and] are doing it to create your own model for your own purposes,” a person close to OpenAI said.
An OpenAI spokesperson added in a statement that the company is aware that China-based firms, along with others, are “constantly trying to distill the models of leading U.S. AI companies.”
“As the leading builder of AI, we engage in countermeasures to protect our IP, including a careful process for which frontier capabilities to include in released models, and believe as we go forward that it is critically important that we are working closely with the U.S. government to best protect the most capable models from efforts by adversaries and competitors to take U.S. technology,” the spokesperson said.
David Sacks, Donald Trump’s newly appointed AI and crypto czar, suggested that intellectual property theft “is possible” in this case.
“There’s substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models, and I don’t think OpenAI is very happy about this,” Sacks told Fox News.