Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

OpenAI says DeepSeek may have used its AI outputs 'inappropriately' to train new models

Jan 29, 2025 - businessinsider.com
OpenAI is investigating DeepSeek for potentially using its AI outputs inappropriately, specifically probing whether DeepSeek trained its AI models using OpenAI's technology without authorization. OpenAI has stated that it takes "aggressive, proactive countermeasures" to protect its technology and is closely reviewing the situation. Concerns have been raised about groups in China using methods like distillation to replicate advanced US AI models, and there are indications that DeepSeek may have engaged in such practices.

DeepSeek has developed top-performing AI models using less-advanced chips at a fraction of the cost of competitors like OpenAI, Google, and Meta, posing a pricing challenge to OpenAI. The situation has impacted tech stocks, raising questions about investments in AI chips and infrastructure. OpenAI's terms prohibit using its model outputs to create competing products, and David Sacks, the White House's AI and crypto czar, suggested that DeepSeek might have committed intellectual property theft. Microsoft informed OpenAI that its security researchers observed potential data siphoning by individuals possibly linked to DeepSeek.

Key takeaways:

  • OpenAI is investigating DeepSeek for potentially using its AI outputs inappropriately, possibly through a method called distillation.
  • DeepSeek has developed AI models using less-advanced chips at a lower cost compared to US rivals like OpenAI, Google, and Meta.
  • OpenAI's terms prohibit using its AI outputs to develop competing models, and the company is taking aggressive measures to protect its technology.
  • Microsoft observed potential data siphoning by individuals possibly linked to DeepSeek, raising concerns about intellectual property theft.
View Full Article

Comments (0)

Be the first to comment!