New Delhi: OpenAI has warned politicians in the US that Chinese artificial intelligence company DeepSeek is employing unethical methods to engineer its next-generation chatbot. The company, in a memo submitted to the House Select Committee on China, accused DeepSeek of using so-called distillation methods to harvest knowledge out of the top US AI models. According to OpenAI, the move poses a threat to American innovation and national security.
The concerns were created after the R1 model of DeepSeek rapidly came to notice due to its ability to provide high-end features at a reduced price. OpenAI asserts that the Chinese firm has implemented more advanced means to circumvent security and steal the products of US systems. Legislators are currently considering the possibility of stricter measures to ensure that American AI research is not abused, reported by Bloomberg.
Distillation allegations and evasion tactics
As explained in the memo, it is possible to use a single AI model to train on the output of another, effectively replicating its advantages through distillation. OpenAI alleged that it had found hidden schemes of evading monitoring systems through so-called obfuscated methods. The company feels that the accounts associated with DeepSeek used third-party routers to hide their origin in accessing the models.
OpenAI mentioned programmatic tools that were purportedly created to harvest huge amounts of AI responses. It put warning notices on networks of illegal resellers which might have allowed indirect access to its services. The problem is that the company started to make some noise inside their organisation when DeepSeek launched its R1 model last year and allegedly collaborated with Microsoft to investigate potential data abuse.
National security and business risks
OpenAI has cautioned that distillation does not just cost the US companies money, but it may undermine built-in protection. Safety filters might not be transferred to AI capabilities when they are reproduced without adequate controls. The company cited that this would heighten the dangers in such sensitive fields as biology and chemistry.
Republican Representative John Moolenaar termed the situation as an extension of a larger trend in China of copying technology. Similar sentiments have been mirrored by other US officials. White House AI advisor David Sacks had earlier said that extensive evidence indicated DeepSeek condensed the knowledge of American systems.
Another overlapping issue is the US export policies on advanced semiconductors. Following the R1 launch of DeepSeek, government officials investigated the possibility of the company having gained access to restricted chips by using third countries. Lawmakers have reviewed records that Nvidia was offering technical assistance regarding the development of the model of DeepSeek.
Since that time, DeepSeek has made small updates and is said to be building out an agent-based system to compete head-on with US competitors. The conflict highlights the increasing tension between Beijing and Washington concerning AI leadership. With both the countries competing to be the top player in the sector, issues of intellectual property and just competition are most likely to be heightened.