OpenAI claims DeepSeek used its model to create ChatGPT rival
What's the story
OpenAI, one of the biggest players in the artificial intelligence (AI) industry, has leveled serious allegations against DeepSeek, the new AI start-up from China. The company claims to have evidence that the Chinese AI firm used its unique model to develop a rival product, according to Financial Times. The accusation raises significant concerns about intellectual property rights and competition within the fast-paced AI sector.
Terms breach
Alleged violation of terms of service
The allegations suggest DeepSeek could have breached OpenAI's terms of service, possibly jeopardizing the integrity of the AI development ecosystem. The accusation comes amid growing scrutiny on AI companies regarding their data usage practices, particularly with respect to international competitors. It remains unclear how these allegations would affect OpenAI's relationships with other AI firms and the potential regulatory implications.
Scenario
Microsoft investigates potential misuse of OpenAI data by DeepSeek
Microsoft is also investigating a possible case of improper data acquisition involving DeepSeek, according to Bloomberg. The tech giant's security researchers have flagged people possibly linked to the Chinese AI firm, who were spotted extracting a large chunk of data using OpenAI's API late last year. Microsoft, which is not just a technology partner to OpenAI but also its biggest investor, has warned the latter about this suspicious activity.
More developments
DeepSeek used OpenAI's models to create own tech: US official
Separately, David Sacks, President Donald Trump's 'AI czar,' has indicated that there's "substantial evidence" DeepSeek used the output from OpenAI's models to create its own tech. He explained a technique called distillation, where one AI model uses another's outputs for training to acquire similar capabilities. "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," Sacks told Fox News.
IP protection
OpenAI acknowledges attempts to distill its models
Addressing Sacks's remarks, an OpenAI spokesperson admitted that companies based in the People's Republic of China and elsewhere are constantly attempting to distill their models. The spokesperson reiterated OpenAI's commitment to safeguarding its intellectual property (IP) and collaborating closely with the US government. "As the leading builder of AI, we engage in countermeasures to protect our IP," the spokesperson said in a statement to Bloomberg.
Market disruption
DeepSeek's new AI model R1 challenges US market leaders
DeepSeek recently launched its open-source AI model called R1, aimed at mimicking human reasoning. The new model has taken on the dominance of US-based companies such as OpenAI, Google, and Meta in the AI space, while being significantly cheaper and more energy efficient. According to DeepSeek, R1 matches or exceeds the performance of leading US developers' products on multiple industry benchmarks, including mathematical tasks and general knowledge.