Chinese AI startup DeepSeek has shaken the global AI industry with its latest advancements, sparking excitement and concern among competitors. The Hangzhou-based company recently unveiled two groundbreaking models: DeepSeek V3, a 671 billion-parameter “mixture of experts” model, and DeepSeek R1, an advanced reasoning system.

DeepSeek’s V3 model “tops the leaderboard among open-source models and rivals the most advanced closed-source models globally,” according to its creators. The company attributes its success to cost efficiency and technological advancements, claiming that V3 was developed with just $6 million in computing power, far below the billions reportedly spent by U.S. firms like OpenAI.

DeepSeek’s competitive edge lies in its affordability. Its R1 model charges only $0.55 per million input tokens, significantly undercutting OpenAI’s $15 per million tokens. Notably, DeepSeek’s models are open-source, licensed under the MIT License. This openness has garnered praise from developers and analysts for promoting innovation but has also raised questions about security and ethical implications.

Meanwhile, U.S. tech stocks have faced turbulence, with Nvidia shares dropping nearly 17% and the Nasdaq losing over 3% amid concerns about DeepSeek’s impact on the industry.

DeepSeek’s privacy policy raises concern

DeepSeek’s data collection and usage practices have raised significant privacy concerns due to the extensive scope of personal and usage information it gathers. According to DeepSeek’s privacy policy, the app collects various forms of user data, including profile information and data/files that have been provided as inputs to the platform, among others.

Also, it clearly states that the data is stored on servers in China and may be transferred internationally, which could be concerning for users in jurisdictions with stricter data protection laws.