Concerns over a potential IP breach are growing as US-based OpenAI looks into whether Chinese AI start-up DeepSeek harvested large amounts of data from ChatGPT-4 to train its model.
DeepSeek’s breakthrough ChatGPT-like AI model made headlines earlier this week. As its R1 app shot to the top of the download charts, it triggered shockwaves in the stock market.
It also shook the tech industry with claims that it developed its R1 model at a fraction of the cost of its rivals.
There has now been accusations that DeepSeek used a technique known as ‘distillation’ from OpenAI’s ChatGPT to build its rival model.
In a video interview with Fox News, White House AI and crypto ‘czar’ David Sacks said: “There’s a technique in AI called distillation ... when one model learns from another model [and] kind of sucks the knowledge out of the parent model.
“And there’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI models.”
While distillation is a common practice...