The Chinese AI chatbot DeepSeek reported a cyberattack yesterday, which prevented new users from registering, although existing users were unaffected. Described as a “widespread, malicious attack,” details on the execution or perpetrators were not disclosed.
The issue was resolved within hours. Recently, DeepSeek has stirred the chip industry and financial markets by claiming to develop a chatbot similar to Western alternatives for a fraction of the cost.
While companies like Google and OpenAI spent tens of millions on their models, DeepSeek reportedly required less than 6 million euros. This implies the algorithm used fewer expensive chips, presenting a breakthrough in AI affordability.
The launch is seen as a “Sputnik moment” by tech investor Marc Andreessen, referencing the surprise caused by the Soviet Union’s satellite launch. OpenAI CEO Sam Altman acknowledged DeepSeek’s impressive results at its price point.
These claims have led to a decline in stock prices for chip industry companies, as an efficient use of chips by DeepSeek could impact their revenues. Dutch companies such as ASML and ASM International saw a 10% drop in Amsterdam, with American Nvidia losing 500 billion in market value.
DeepSeek, available in Apple and Google app stores since early this year, is now the most downloaded free app at Apple. The chatbot does not require a subscription, unlike ChatGPT.
However, experts advise caution with sensitive data, as the handling of user information by companies remains unclear. Additionally, the Chinese app appears to be restricted by communist censorship, evidenced by its inability to provide information on the 1989 student protests when queried by NOS.
Source: NOS