The sudden arrival of China’s DeepSeek AI chatbot has thrown the US’s AI industry into what Forbes calls a “panic.”
AI-related companies’ stocks fell sharply on Monday, with chipmaker Nvidia leading the way to a 17% stock price drop, wiping out nearly USD $600 billion of the company’s market cap. (A rebound on Tuesday clawed back some of those losses.)
The reason for the panic? DeepSeek appears to have created a chatbot with similar capabilities to those of US-made chatbots, but using a fraction of the resources and capital needed to develop the US models. The company says it spent just $5.6 million developing the chatbot, compared to the more than $100 million that OpenAI spent developing ChatGPT-4.
Now it appears that DeepSeek may have accomplished this through intellectual property theft. According to news reports, OpenAI – maker of the ChatGPT chatbot that triggered the global AI craze a few years ago – is investigating whether DeepSeek violated its intellectual property in creating its R1 artificial intelligence model.
OpenAI was reportedly notified of the possible violation by its key investor, Microsoft, according to Bloomberg, which first reported on the matter. The AI company then blocked DeepSeek’s access to ChatGPT, according to the Financial Times.
For those following the growing battle between copyright owners and AI developers, this seems like a clear case of tables turning: OpenAI has been dragged into court by the likes of The New York Times, comedian-writer Sarah Silverman, Game of Thrones author George R.R. Martin, and German music licensing organization GEMA. They all accuse the company of training its AI on copyrighted content without permission.
In an interview with Fox News, President Donald Trump’s artificial intelligence czar, David Sacks, said there’s “substantial evidence” that DeepSeek used “distillation” to develop its AI technology.
Distillation is a process in which a smaller, more efficient AI model is trained to mimic the outputs of a larger, less efficient model in order to replicate its behavior. The technique is used to create AI services that are cheaper to develop, and require less processing power and energy to run.
However, in this instance, it appears DeepSeek used ChatGPT-4 to train its own, smaller model, a violation of OpenAI’s terms of service, FT reported.
“The issue is when you [take it out of the platform and] are doing it to create your own model for your own purposes,” an unnamed individual close to OpenAI said.
Some experts say it could be hard to stop AI developers from using distillation to piggyback on the achievements of other AI developers, because the practice is widespread.
“It is a very common practice for start-ups and academics to use outputs from human-aligned commercial LLMs, like ChatGPT, to train another model,” Ritwik Gupta, a PhD candidate in AI at the University of California, Berkeley, told FT.
“It is not surprising to me that DeepSeek supposedly would be doing the same. If they were, stopping this practice precisely may be difficult.”
“We know [People’s Republic of China]-based companies – and others – are constantly trying to distill the models of leading US AI companies,” OpenAI said in a statement.
“We engage in countermeasures to protect our IP, including a careful process for which frontier capabilities to include in released models, and believe . . . 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.”
That marks a striking change of tone from the one OpenAI has taken when defending itself against accusations of IP theft. In response to the suit brought by Sarah Silverman and other authors, OpenAI indicated it plans to defend itself by arguing that using copyrighted works to train AI should be considered a “fair use” exemption to US copyright laws.
The “fair use” argument has been made by other AI developers, notably Anthropic (sued by Universal Music Group (UMG), Concord, and ABKCO for allegedly violating copyrights on lyrics) and Suno and Udio, two AI music-making apps, sued by the three music majors, that are banking so heavily on the fair-use argument that they all but admitted to using UMG, Warner Music, and Sony Music content without permission in developing their apps.
In the wake of DeepSeek’s release of R1, and the ensuing market chaos, AI developers may well be rethinking their liberal stance on intellectual property.Music Business Worldwide
Article by:Source:
