After cyberattacks: DeepSeek deactivates registration
DeepSeek reports large-scale, malicious attacks on the services. Performance suffers as a result. Registrations are temporarily deactivated.
(Image: Erstellt mit KI in Bing Designer durch heise online / dmk)
The hullabaloo surrounding the AI company DeepSeek has further repercussions: The company reports cyberattacks against its services. New registrations are therefore currently not possible, and performance is also suffering.
(Image: Screenshot / dmk)
DeepSeek writes on its status website that it is currently experiencing “reduced performance”. The attacks began on Monday, January 27th. Initially, neither the log-in nor the API or web interface could be used, but after initial investigations, the services became available again – but with reduced speed.
New registration deactivated
On Tuesday of this week, DeepSeek deactivated registration, but existing accounts could be used again. In the evening, the company was apparently able to initiate and implement countermeasures. DeepSeek is currently monitoring their effectiveness. New registrations are currently still deactivated, but only temporarily, as DeepSeek explains.
Videos by heise
The attacks follow the tremor that DeepSeek has caused in the AI industry. Allegedly, the company has achieved AI models with significantly lower costs and hardware for training, which can apparently keep up with those of the previous industry leader OpenAI, for example. The company has even presented an image generator.
Meanwhile, the share prices of AI companies have plummeted, with Nvidia alone suffering losses of around 600 billion US dollars. Previously renowned AI companies have reacted rather panicked to the newcomer from China.
DeepSeek provides the source code for its language models and the Janus image generator free of charge as open source under an MIT license. Interested parties can access the sources for DeepSeek-V3, DeepSeek-VL2, DeepSeek-R1 and Janus and others in DeepSeek's GitHub repositories.
(dmk)
