last month, deepseek made headlines after it claimed its ai model was significantly cheaper to develop than those of its competitors, causing a sharp decline in us tech stock prices. social media users reacted humorously, with one joke stating, “i can’t believe chatgpt lost its job to ai.”
what is deepseek?
deepseek is a chinese ai startup based in hangzhou, founded in 2023 by entrepreneur liang feng, who previously co-founded the $7 billion hedge fund high-flyer in 2016.
in january, deepseek launched its latest ai chatbot, deepseek r1, a free tool similar in appearance and function to openai’s chatgpt. ai chatbots simulate human conversations, providing responses based on internet-sourced and pre-trained information. these programs are widely used for tasks such as solving math problems, drafting documents, and translating or generating code.
how is deepseek different from chatgpt?
deepseek stands out for its significantly lower development costs.
in december, the company published a report stating that training its latest model required less than $6 million worth of nvidia h800 chips—only a fraction of the billions spent by us tech giants like openai and google on their ai models. openai’s ceo, sam altman, previously revealed that training gpt-4 cost more than $100 million.
deepseek r1 was released on january 20 and quickly became the top-ranked app on apple’s app store. by january 27, nvidia’s market capitalization had dropped by nearly $600 billion, with its stock plummeting 17% in response to deepseek’s disruptive pricing model.
countries restricting deepseek
united states
- the wall street journal reported that us lawmakers are planning legislation to ban deepseek from government devices.
- nasa blocked deepseek from its systems on january 31.
- the us navy warned personnel against using the chatbot due to security and ethical concerns.
south korea
- the ministry of trade, industry, and energy has temporarily banned deepseek for government employees.
- the government issued a cautionary statement on ai programs, including chatgpt and deepseek.
- korea hydro & nuclear power, a government-run entity, restricted access to deepseek and similar ai tools.
australia
- the government blocked deepseek on all official devices, citing security risks.
- the department of home affairs ordered the removal of deepseek applications from all government systems.
- home affairs minister tony burke emphasized the decision was made to protect national security.
italy
- the italian data protection authority (garante) restricted deepseek’s processing of italian user data due to privacy concerns.
- authorities are investigating how deepseek collects and handles personal data.
taiwan
- the government prohibited deepseek’s use within government departments, citing security threats.
why are countries concerned?
governments fear that deepseek poses security risks, particularly regarding user data privacy.
deepseek’s privacy policy states that it collects user information, including:
- personal details (email, phone number, date of birth)
- chat history (text or audio input)
- technical data (ip address, keystroke patterns, operating system)
this data is shared with service providers and advertisers and retained for an unspecified period. similarly, openai collects and stores user data but claims to retain it only as long as necessary.
on february 7, abc news cited a report from feroot security, claiming that deepseek has hidden code that can send user data directly to the chinese government via cmpassport.com, a portal linked to state-owned china mobile. al jazeera has not independently verified this claim.
is this practice unusual?
data collection is standard for major tech companies. firms like google, meta, and openai use user data for advertising, algorithm improvements, and ai training. however, they are subject to regulatory scrutiny, particularly in the us and eu.
chinese companies, on the other hand, are viewed differently due to china’s national intelligence law, which could grant the state access to corporate data. western governments fear chinese platforms may be used for surveillance, espionage, or influence operations.
while western tech firms face regulatory hurdles, they are generally not considered national security threats by their own governments. in contrast, chinese platforms often face outright bans.
for example, in 2023, italy temporarily blocked chatgpt over gdpr concerns but lifted the ban after openai addressed regulatory issues. meanwhile, china itself blocks western platforms like facebook, x (twitter), and chatgpt for all users within its borders.