Technical experts in the US are investigating whether data output from OpenAI was improperly obtained by a group allegedly linked to Chinese artificial intelligence (AI) startup DeepSeek.
Major news outlets reported on Tuesday that Microsoft’s security researchers observed that “individuals they believed to be connected to DeepSeek” were involved in an unauthorized data transfer, known as a data exfiltration or distillation.
They said a large amount of data was exfiltrated in the last quarter of 2024 via OpenAI’s application programming interface (API), which they said is the main way that software developers and business customers buy OpenAI’s services.
ALSO SEE: Tech Selloff Extends to Japan as DeepSeek Puts Focus on AI Costs
Microsoft, the largest investor for OpenAI, notified the company of suspicious activity, according to one report.
Low-cost Chinese AI startup DeepSeek, an alternative to major US tech rivals, sparked a tech stock selloff on Monday and Tuesday as its free AI assistant overtook OpenAI’s ChatGPT on Apple’s App Store in the United States.
David Sacks, the White House’s AI and crypto czar, told Fox News in an interview earlier on Tuesday that it was “possible” that DeepSeek stole intellectual property from the United States.
“There’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models,” Sacks said.
The Financial Times reported that OpenAI said it had seen some evidence of “distillation”, which it suspects to be from DeepSeek, and believes that was likely used to train rival model/s. That would violate OpenAI’s terms of service.
Asked for comment on the report, an OpenAI spokesperson echoed Sacks in a statement that noted China-based companies and others were constantly attempting to replicate the models of leading US AI companies, without specifically naming DeepSeek or any other company.
“We engage in counter-measures 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 US government to best protect the most capable models from efforts by adversaries and competitors to take US technology.”
Microsoft declined to comment, while DeepSeek, which is based in Hangzhou, could not be immediately reached for a comment.
The release of DeepSeek’s AI model has caused analysts to question whether the vast amounts of money and computing power that US tech giants are throwing at AI, such as the Stargate project by OpenAI and others to build giant data centres, is necessary.
Gil Luria, head of tech research at investment group DA Davidson, told CNN: “DeepSeek makes it very clear that that the current trajectory of scaling up of data centres is highly unlikely to be economic to Nvidia’s customers.”
US Navy personnel ordered to avoid AI app
Meanwhile, the White House has said the National Security Council is looking into potential security implications of DeepSeek’s alleged AI breakthroughs, according to a report to Axios
But it said that beyond the economic impacts on some of America’s most valuable tech companies, “it’s unclear what precisely the national security ramifications might be.”
And, the US Navy has warned its personnel to avoid use of DeepSeek’s AI model for work tasks or personal use, due to “potential security and ethical concerns associated with the model’s origin and usage,” according to a report by CNBC.
Australian cybersecurity executives and the federal opposition also issued a warning, the Sydney Morning Herald reported on Wednesday.
It quoted a cybersecurity expert, Alistair MacGibbon, as saying: “Anything we enter into DeepSeek is going straight into the hands of the Chinese Communist Party.”
Spotlight on founder
The booming interest in DeepSeek has put the spotlight on founder Liang Wenfeng, who participated in a symposium hosted by Chinese PM Li Qiang recently that drew industry experts from technology, education, science, culture, health and sports.
Beijing has made AI a national priority and the appearance of Liang, 40, at such an event showed his growing prominence in the industry, a recent report by the South China Morning Post said.
It said Liang studied at Zhejiang University in Hangzhou and explored how AI could be used to automate stock investments, which led to him founding High-Flyer Quant, one of the country’s largest quant hedge funds. That provided funds to build AI infrastructure and conduct large-scale research, which led to him spinning off DeepSeek in mid-2023 and launching a series of AI models used by developers and the creation of its own chatbot.
“High-Flyer Quant managed to buy more than 10,000 Nvidia graphics processing units before the US government imposed AI chip restrictions on China, it said, citing local media outlet 36kr. “On its website, the hedge-fund manager said it spent 200 million yuan and 1 billion yuan in 2020 and 2021, respectively, to build its Fire-Flyer series of AI computing clusters.”
China’s AI market is expected to be worth 5.6 trillion yuan ($765 billion) by 2030, according to state-backed investment vehicle China International Capital Corp (CICC).
DeepSeek said on Monday it would temporarily limit user registrations because of “large-scale malicious attacks” on its services, before later resuming operations.
Liang Wenfeng has, not surprisingly, been described as the “biggest dark horse” in the vibrant AI sector this year.
CNN said that in a rare interview with local media outlet 36Kr last July, Liang said: “We often say there’s a one or two-year gap between China and the US, but the real gap is between originality and imitation. If this doesn’t change, China will always be a follower.”
- Reuters with extensive additional inputs and editing by Jim Pollard
NOTE: Many additional details were added to this report on January 29, 2025.
ALSO SEE:
‘AI Breakthrough’ by China’s DeepSeek Rocks US Tech Giants
SoftBank, UAE’s MGX Join $500bn Data Centre Deal With OpenAI
Bytedance ‘Plans $20 Billion Outlay on AI Infrastructure This Year’
Biden Restricts Access to AI Chips to US Firms And Its Allies
Biden’s Final Fling: Curbs on US-Made AI Chips, Russia Sanctions
US Rules to Limit Investment in China’s Chips, Quantum and AI
US Ban on Investment Not Good For AI Sector, China Says
China’s Bid to ‘Cheat’ a Way to Chip Prominence is Failing: Envoy
US Plans New Rules for Cloud Firms to Cut Off China AI Access
Venture Capitalists’ Backing China Put Democracy at Risk – NYP
Nvidia to Stop Some AI Chip Exports to China Immediately
China’s Military, AI Bodies Still Buying Nvidia Chips Despite US Ban
US Releases Detailed Rules For Export Curbs on AI Chips to China