The New DeepSeek R1 AI Platform

What makes R1 most interesting is that, unlike other top models from tech giants, it’s open source, meaning anyone can download and use it. That said, DeepSeek has not disclosed R1’s training dataset. DeepSeek is also cheaper than comparable US models. For reference, R1 API access starts at $0.14 for a million tokens, which is a fraction of the $7.50 that OpenAI charges for the equivalent tier. One drawback that could impact its long-term competition with o1 and other US-made models is censorship. Chinese models often include blocks on certain subject matter, meaning that while they function comparably to other models, they may not answer some queries. Another problem is privacy concerns. “The personal information we collect from you may be stored on a server located outside of the country where you live,” DeepSeek’s privacy policy states. “We store the information we collect in secure servers located in the People’s Republic of China.” The policy outlines that DeepSeek collects plenty of information, including but not limited to: • IP address, unique device identifiers, and cookies • Date of birth (where applicable), username, email address and/or telephone number, and password • Your text or audio input, prompt, uploaded files, feedback, chat history, or other content that you provide to our model and Services • Proof of identity or age, feedback or inquiries about your use of the Service, if you contact DeepSeek. DeepSeek has also been proven to be considerably less secure than it’s major US counterparts, like the latest version of ChatGPT. It is considerably easier to “jailbreak” DeepSeek, which means trick it into giving answers to questions that could aid in criminal activities.

Update 2025-01-30

Researchers at Wiz has published information that the database that is used by DeepSeek is compeletely open and unauthenticated, exposing data. According to Wiz, this database contained a significant volume of chat history, backend data and sensitive information, including log streams, API Secrets, and operational details. To make matters worse, Wiz said, the exposure allowed for full control of the database and potential privilege escalation within the DeepSeek environment, without any authentication or barrier to external access.

Assessment

The release of DeepSeek comes at a strategic time. Large tech giants have invested heavily in new AI models, but are now struggling to monetize their investments. Even before the release of DeepSeek there has been discussions about AI companies being overvalued. DeepSeek is funded by a Chinese tech giant, which in practice means it likely has access to government funding, so it doesn’t have the same pressure to produce return on investment. It’s possible that by releasing DeepSeek at a far lower cost than it’s US counterparts, China is trying to harm the US AI industry by hurting their revenue. Users also need to be aware that any data shared with the DeepSeek platform could be subject to government access under China’s cybersecurity laws, which mandate that companies must provide access to data upon request by authorities. It is best to assume that all information fed into DeepSeek, together with a lot of personal information about the user, will in effect become available to the Chinese authorities and intelligence services, and write policies based on that assumption. The fact that DeepSeek has an unprotected database and is easier to jailbreak, means that it’s highly likely also less robust in resisting various other attacks that can potentially harm the user or manipulate the output of the AI.

References

https://www.zdnet.com/article/what-to-know-about-deepseek-ai-from-cost-claims-to-data-privacy/
https://www.kelacyber.com/blog/deepseek-r1-security-flaws/
https://x.com/attorneyjeremy1/status/1884428727204872681
https://www.theregister.com/2025/01/30/deepseek_database_left_open/