The people behind ChatGPT have expressed their suspicion that China’s ultra cheap DeepSeek AI models were built upon OpenAI data.
This week, Donald Trump said DeepSeek should be considered a “wake-up call” for the U.S. tech industry after Nvidia saw almost $600 billion wiped off its market value.
The emergence of DeepSeek sent stocks in companies heavily invested in artificial intelligence into freefall. Nvidia, which dominates the market for GPUs upon which AI models run, was hit hardest when its shares tumbled 16.86% — the biggest loss in Wall Street history.
Microsoft, Meta Platforms, and Google parent Alphabet fell between 2.1% and 4.2%, while AI server maker Dell Technologies was down 8.7%.
DeepSeek claims its R1 model is a significantly cheaper alternative to western offerings such as ChatGPT. It’s built on the open source DeepSeek-V3, which reportedly requires far less computing power than western models and is estimated to have been trained for just $6 million.
While some have disputed this claim, DeepSeek has had the effect of calling into question the billions American tech companies are investing in AI, which in turn has spooked investors. Indeed, DeepSeek shot to the top of the most downloaded free app chart in the U.S. off the back of rising discussion about its effectiveness.
Now, Bloomberg has reported that OpenAI and Microsoft are looking into whether DeepSeek used OpenAI’s API to integrate OpenAI’s AI models into DeepSeek’s own models. “We know PRC (China) based companies — and others — are constantly trying to distill the models of leading U.S. AI companies,” OpenAI told Bloomberg.
Distillation is a technique developers use to train AI models by extracting data from larger, more capable ones. It is a violation of OpenAI’s terms of service.
“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,” OpenAI continued.
President Donald Trump’s artificial intelligence czar David Sacks told Fox News: “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. I think one of the things you’re going to see over the next few months is our leading AI companies taking steps to try to prevent distillation.”
Observers were quick to point out the irony here given OpenAI is accused of ripping off the internet in the creation of Chat GPT. Tech PR and writer Ed Zitron tweeted: “I’m so sorry I can’t stop laughing. OpenAI, the company built on stealing literally the entire internet, is crying because DeepSeek may have trained on the outputs from ChatGPT. They’re crying their eyes out. What a bunch of hypocritical little babies.”
In January 2024, OpenAI insisted it was “impossible” to create AI tools like ChatGPT without copyrighted material.
In a submission to the UK’s House of Lords communications and digital select committee, OpenAI said it could not train large language models such as ChatGPT without access to copyrighted work.
“Because copyright today covers virtually every sort of human expression — including blogposts, photographs, forum posts, scraps of software code, and government documents — it would be impossible to train today’s leading AI models without using copyrighted materials,” OpenAI said in its submission, as reported by paywalled UK newspaper the Telegraph.
“Limiting training data to public domain books and drawings created more than a century ago might yield an interesting experiment, but would not provide AI systems that meet the needs of today’s citizens,” OpenAI added.
Training AI models on copyrighted materials has become one of the technology industry’s hottest topics as generative AI has exploded in recent years. In December 2023, the New York Times sued OpenAI and investor Microsoft for the “unlawful use” of its work to create their products. In response, OpenAI said it believes training is “fair use,” and insisted: “We support journalism, partner with news organizations, and believe The New York Times lawsuit is without merit.”
The New York Times’ lawsuit followed a lawsuit in September 2023 brought by 17 authors, including Game of Thrones’ George R. R. Martin, who allege “systematic theft on a mass scale.”
In August that year, District Judge Beryl Howell upheld a U.S. Copyright Office finding that AI art could not be copyrighted. This finding dates back to 2018, when the Copyright Office claimed “the nexus between the human mind and creative expression” is critical to the grounds of copyright protection.
Image credit: Andrey Rudakov/Bloomberg via Getty Images.
Wesley is the UK News Editor for IGN. Find him on Twitter at @wyp100. You can reach Wesley at [email protected] or confidentially at [email protected].
Add comment