OpenAI, the creator of ChatGPT, has raised concerns that competitors, including those in China, are leveraging its technology to make significant strides in developing their own AI tools.
This week, the rise of DeepSeek, a Chinese app capable of mimicking ChatGPT’s performance, has shaken the dominance of OpenAI and other US-based companies in the AI field. DeepSeek’s ability to replicate ChatGPT at what appears to be a fraction of the cost has raised alarm bells.
According to Bloomberg, Microsoft, a major investor in OpenAI, is looking into whether OpenAI’s data has been used without authorization.
The BBC has reached out to both Microsoft and DeepSeek for comment.
OpenAI’s concerns about potential misuse of its models have been echoed by David Sacks, the newly appointed White House “AI and crypto czar.” In a Fox News interview, Sacks suggested that DeepSeek may have employed a technique called knowledge distillation to improve its app’s capabilities, essentially extracting knowledge from OpenAI’s models to enhance its own.
“There’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models,” Sacks said. He predicted that over the coming months, leading US AI companies would likely take steps to prevent this kind of “distillation,” which could slow the development of competing models.
OpenAI itself has confirmed that Chinese and other companies are continuously trying to “distill” the models from leading US AI firms. The organization emphasized the need for close collaboration with the US government to safeguard the most advanced models.
What is DeepSeek?
DeepSeek is a Chinese-developed AI app that appears to replicate the functionality of ChatGPT at a significantly lower cost. There are concerns regarding how DeepSeek was able to achieve this, with some suggesting it may have used models from companies like OpenAI without authorization.
Is DeepSeek safe to use?
Given the controversy surrounding how DeepSeek has developed its AI models, there are questions about whether it is ethical or legal to use. If it has indeed employed unauthorized use of data or models, there may be legal and privacy concerns for users.
DeepSeek vs. ChatGPT: How do they compare?
While DeepSeek has made waves by offering similar services to ChatGPT, questions linger about whether it has truly developed its technology independently. Experts are skeptical of claims that DeepSeek trained its models at a fraction of the cost, suggesting that distillation from existing models may have played a key role in its development.
‘Deceptive’ claims
Naomi Haefner, an assistant professor of technology management at the University of St. Gallen in Switzerland, pointed out that the suggestion that DeepSeek trained its models “cheaply” might be misleading. “It is unclear whether DeepSeek really trained its models from scratch,” she said, adding that OpenAI believes DeepSeek may have misappropriated significant amounts of data.
Crystal van Oosterom, AI Venture Partner at OpenOcean, agreed that DeepSeek had “clearly built upon publicly available research from major American and European institutions and companies.” However, the line between “building on” existing research and infringing on intellectual property rights is murky. This issue is particularly relevant in the AI industry, where accusations of intellectual property violations have been a recurring theme for major US companies.
US officials are increasingly concerned about the national security implications of DeepSeek’s rapid rise, according to White House press secretary Karoline Leavitt. “I spoke with [the National Security Council] this morning, and they are looking into what [the national security implications] may be,” Leavitt said. She also reiterated US President Donald Trump’s comments from the previous day, urging that DeepSeek should serve as a wake-up call for the US tech industry.
This heightened scrutiny comes after reports that the US Navy has banned its personnel from using DeepSeek’s app due to potential security and ethical concerns. CNBC reported that the Navy sent an email to its members warning against using the app, citing concerns about the model’s origin and usage. The decision reflects ongoing fears about how DeepSeek’s technology may be leveraged and the potential risks it poses to national security.
Data safety experts are urging caution for users of DeepSeek, citing concerns about the app’s collection of vast amounts of personal data, which is stored on servers in China. This raises significant privacy and security risks, particularly given the potential for unauthorized access to sensitive information.
In the midst of these concerns, DeepSeek has claimed it has been the target of cyberattacks. On Monday, the company announced it would temporarily limit new registrations due to “large-scale malicious attacks” on its software. A banner on the company’s website currently notifies users that registration may be delayed as a result of these ongoing security challenges. This adds another layer of uncertainty regarding the app’s safety and reliability.
