DeepSeek is a Chinese AI firm based in Hangzhou, China that was founded in May 2023. The organization was introduced by Liang Wenfeng, who is the co-founder of High-Flyer, a China-based quantitative hedge fund. This independent AI research aims to develop open-source LLMs. Since its inception, it has undergone several iterations and built out several distinct variations.
DeepSeek and Open AI are different from each other in several aspects, and is briefed in the table below:
| DeepSeek | Open AI | |
| Development focus | Efficient, open-source models | Broad AI capabilities |
| Key models | DeepSeek – V3,DeepSeek – R1 | GPT – 4o, o1 |
| Specialized models | DeepSeek Coder (coding),Janus Pro (vision model) | Dall – E (image generation), Whisper (speech recognition) |
| API pricing (per million tokens) | DeepSeek – R1: $0.55 (input), $2.19 (output) | O1: $15 (input), $60 (output) |
| Open source policy | Mostly open source | Limited |
| Training approach | Reinforcement learning | Supervised and instruction-based fine-tuning |
| Development cost | Less than $6 million for DeepSeek – R1, according to the company | Hundreds of millions of dollars for o1 (estimated) |
What Data Does Deepseek Collect?
As per DeepSeek’s privacy policy, it collects a significant amount of personal information from users, and it is stored in the ‘secure servers’ in China. This information is used for improving security, safety, and stability and kept with the service providers, corporate group, and advertising partners as long as necessary. The collected information may include:
- Email address, phone number, and date of birth entered during the creation of the account.
- Text, audio, and chat history are entered by the user.
- Includes technical information such as the phone’s model and operating system and ‘keystroke patterns.’
There are genuine concerns about the technological potential of DeepSeek. Concerns are raised about privacy and potential misuse of user data. Experts advise users to exercise caution whenever they are using the chatbot and be of utmost vigilance when sharing sensitive information. There are still some unanswered questions regarding how DeepSeek manages data and privacy.
Since Chinese tech companies are linked to the Chinese government, there is the possibility of exploitation of user data for intelligence purposes. This concern was raised by the Western government. The first Western government official questioned about DeepSeek’s privacy policy was Australia’s science minister, Ed Husic. Besides the Western government, the United States also raised serious concerns about DeepSeek. National Security Council investigated the potential risk related to the chatbot. Furthermore, the personnel are restricted from using DeepSeek.
The launch of a new AI assistant is giving tough competition to DeepSeek, which claims that it only uses less data at a fraction of the cost of other services in comparison. What is happening behind DeepSeek is it creates a digital profile of persons with every keystroke, tracked interaction, and, most importantly, the data the person has taken from third-party sources. This can push you to more serious security and privacy concerns than ChatGPT’s.
How Dangerous Is The Ai-Driven Profiling?
This can be a very serious concern as more data are collected by DeepSeek, and those data will be used for several cases;
- They may use the data to conduct behavioral tracking and thereby present the search results even before you search it.
- There is no assurance about the persons who have access to the data stored on servers in China.
- High risk of tracking user sentiment to identify personal biases.
All these point towards privacy as it could track a person’s behavior and this can be used for several cases, such as exploitation for manipulation.
Conclusion
The introduction of artificial intelligence tools like DeepSeek has indeed made powerful innovations. However, the risks associated with these innovations cannot be neglected. Individuals using these tools should take action to protect their information from being manipulated. Implementation of certain actions could help you to safeguard your data, including not sharing sensitive or personal information such as passwords and work-related information, using a Virtual Private Network (VPN) to mask your digital footprint, and if there is any provision for disabling data sharing make use of it, and demand for higher privacy standards. Always stay informed and mindful whenever dealing with AI tools.
