What to know about DeepSeek AI, from cost claims to data privacy

A new player has entered the AI villa, and it’s creating significant disruption. 

Chinese AI startup DeepSeek made waves last week when it released the full version of R1, the company’s open-source reasoning model that can outperform OpenAI’s o1. On Monday, App Store downloads of DeepSeek’s AI assistant — which runs V3, a model DeepSeek released in December — topped ChatGPT, which had previously been the most downloaded free app. 

Also: I tested DeepSeek’s R1 and V3 coding skills – and we’re not all doomed (yet)

DeepSeek R1 has also already climbed to the third spot overall on HuggingFace’s Chatbot Arena, under several Gemini models as well as ChatGPT-4o. Almost as soon as it dethroned OpenAI, DeepSeek began limiting signups due to a supposed cyberattack, and then followed up its big splash with a promising new image model. 

What is DeepSeek?

Founded by Liang Wenfeng in May 2023 (and thus not even two years old), the Chinese startup has challenged established AI companies with its open-source approach. According to Forbes, DeepSeek’s edge may lie in the fact that it is funded only by High-Flyer, a hedge fund also run by Wenfeng, which gives the company a funding model that supports fast growth and research. 

What is DeepSeek R1?

Released in full last week, R1 is DeepSeek’s flagship reasoning model, which performs at or above OpenAI’s lauded o1 model on several math, coding, and reasoning benchmarks. 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. So far, all other models it has released are also open-source. 

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. 

Also: Enterprises are hitting a ‘speed limit’ in deploying Gen AI – here’s why

DeepSeek claims in a company research paper that its V3 model cost $5.6 million to train, a number that is being circulated (and disputed) as the entire development cost of the model. As the AP reported, some lab experts believe the paper is referring to only the final training run for V3, not its entire development cost (which would be a fraction of what tech giants have spent to build competitive models). Some suggest DeepSeek’s costs don’t include earlier infrastructure, R&D, data, and personnel costs.

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 (see how DeepSeek’s AI assistant responds to queries about Tiananmen Square and Taiwan here). 

In December, ZDNET’s Tiernan Ray compared R1-Lite’s ability to explain its chain of thought to that of o1, and the results were mixed. 

Also: Enterprises are hitting a ‘speed limit’ in deploying Gen AI – here’s why

Of course, all popular models come with their own red-teaming background, community guidelines, and content guardrails — but at least at this stage, American-made chatbots are unlikely to refrain from answering queries about historical events. 

Privacy concerns

Data privacy worries that have circulated around TikTok — the Chinese-owned social media app that is now somewhat banned in the US — are also cropping up about DeepSeek. 

“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.”

Also: ‘Humanity’s Last Exam’ benchmark is stumping top AI models – can you do any better?

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

The policy continues: “Where we transfer any personal information out of the country where you live, including for one or more of the purposes as set out in this Policy, we will do so in accordance with the requirements of applicable data protection laws.” It does not mention GDPR compliance.

“Users need to be aware that any data shared with the platform could be subject to government access under China’s cybersecurity laws, which mandate that companies provide access to data upon request by authorities,” said Adrianus Warmenhoven, a member of NordVPN‘s security advisory board, told ZDNET via email.

“DeepSeek’s AI model has faced growing backlash for its refusal to address political topics,” he added. “This has sparked concerns about potential biases and external influence on the platform’s content moderation policies.”

Also: How to protect your privacy from Facebook – and what doesn’t work

According to some observers, the fact that R1 is open-source means increased transparency, giving users the opportunity to inspect the model’s source code for signs of privacy-related activity. Regardless, DeepSeek also released smaller versions of R1, which can be downloaded and run locally to avoid any concerns about data being sent back to the company (as opposed to accessing the chatbot online). All chatbots, including ChatGPT, are collecting some degree of user data when queried via the browser. 

What this means for AI at large 

R1’s success highlights a sea change in AI that could empower smaller labs and researchers to create competitive models and diversify the field of available options. For example, organizations without the funding or staff of OpenAI can download R1 and fine-tune it to compete with models like o1. Just before R1’s release, researchers at UC Berkeley created an open-source model that is on par with o1-preview, an early version of o1, in just 19 hours and for roughly $450. 

Given how exhorbitant AI investment has become, many are speculating that this development could burst the AI bubble. Multiple reports indicate the stock market is already panicking. 

Also: $450 and 19 hours is all it takes to rival OpenAI’s o1-preview

DeepSeek’s ascent comes at a critical time for Chinese-American tech relations, just days after the long-fought TikTok ban went into partial effect. Ironically, DeepSeek lays out in plain language the fodder for security concerns that the US struggled to prove about TikTok on its prolonged effort to enact a ban.